{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 19.176987614862167,
  "eval_steps": 500,
  "global_step": 48000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00039952057530962844,
      "grad_norm": 7.03125,
      "learning_rate": 0.0005,
      "loss": 3.7958,
      "step": 1
    },
    {
      "epoch": 0.0007990411506192569,
      "grad_norm": 47.5,
      "learning_rate": 0.0005,
      "loss": 16.0438,
      "step": 2
    },
    {
      "epoch": 0.0011985617259288853,
      "grad_norm": 41.5,
      "learning_rate": 0.0005,
      "loss": 11.6589,
      "step": 3
    },
    {
      "epoch": 0.0015980823012385138,
      "grad_norm": 40.0,
      "learning_rate": 0.0005,
      "loss": 12.8325,
      "step": 4
    },
    {
      "epoch": 0.001997602876548142,
      "grad_norm": 36.75,
      "learning_rate": 0.0005,
      "loss": 12.5015,
      "step": 5
    },
    {
      "epoch": 0.0023971234518577705,
      "grad_norm": 21.625,
      "learning_rate": 0.0005,
      "loss": 7.0332,
      "step": 6
    },
    {
      "epoch": 0.002796644027167399,
      "grad_norm": 13.8125,
      "learning_rate": 0.0005,
      "loss": 4.1196,
      "step": 7
    },
    {
      "epoch": 0.0031961646024770275,
      "grad_norm": 6.5,
      "learning_rate": 0.0005,
      "loss": 2.8851,
      "step": 8
    },
    {
      "epoch": 0.003595685177786656,
      "grad_norm": 4.5,
      "learning_rate": 0.0005,
      "loss": 2.5168,
      "step": 9
    },
    {
      "epoch": 0.003995205753096284,
      "grad_norm": 8.1875,
      "learning_rate": 0.0005,
      "loss": 2.9302,
      "step": 10
    },
    {
      "epoch": 0.0043947263284059125,
      "grad_norm": 5.1875,
      "learning_rate": 0.0005,
      "loss": 2.3453,
      "step": 11
    },
    {
      "epoch": 0.004794246903715541,
      "grad_norm": 3.5,
      "learning_rate": 0.0005,
      "loss": 2.1391,
      "step": 12
    },
    {
      "epoch": 0.0051937674790251695,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0005,
      "loss": 1.9119,
      "step": 13
    },
    {
      "epoch": 0.005593288054334798,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005,
      "loss": 1.8872,
      "step": 14
    },
    {
      "epoch": 0.0059928086296444265,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0005,
      "loss": 1.8285,
      "step": 15
    },
    {
      "epoch": 0.006392329204954055,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0005,
      "loss": 1.8244,
      "step": 16
    },
    {
      "epoch": 0.0067918497802636835,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0005,
      "loss": 1.8067,
      "step": 17
    },
    {
      "epoch": 0.007191370355573312,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005,
      "loss": 1.7836,
      "step": 18
    },
    {
      "epoch": 0.0075908909308829405,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0005,
      "loss": 1.772,
      "step": 19
    },
    {
      "epoch": 0.007990411506192568,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0005,
      "loss": 1.7875,
      "step": 20
    },
    {
      "epoch": 0.008389932081502197,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0005,
      "loss": 1.7794,
      "step": 21
    },
    {
      "epoch": 0.008789452656811825,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0005,
      "loss": 1.7714,
      "step": 22
    },
    {
      "epoch": 0.009188973232121454,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0005,
      "loss": 1.7494,
      "step": 23
    },
    {
      "epoch": 0.009588493807431082,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0005,
      "loss": 1.7309,
      "step": 24
    },
    {
      "epoch": 0.00998801438274071,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005,
      "loss": 1.7147,
      "step": 25
    },
    {
      "epoch": 0.010387534958050339,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005,
      "loss": 1.7107,
      "step": 26
    },
    {
      "epoch": 0.010787055533359968,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0005,
      "loss": 1.7006,
      "step": 27
    },
    {
      "epoch": 0.011186576108669596,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0005,
      "loss": 1.6963,
      "step": 28
    },
    {
      "epoch": 0.011586096683979225,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0005,
      "loss": 1.6782,
      "step": 29
    },
    {
      "epoch": 0.011985617259288853,
      "grad_norm": 0.5625,
      "learning_rate": 0.0005,
      "loss": 1.6598,
      "step": 30
    },
    {
      "epoch": 0.012385137834598482,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0005,
      "loss": 1.646,
      "step": 31
    },
    {
      "epoch": 0.01278465840990811,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0005,
      "loss": 1.6078,
      "step": 32
    },
    {
      "epoch": 0.013184178985217739,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005,
      "loss": 1.5961,
      "step": 33
    },
    {
      "epoch": 0.013583699560527367,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0005,
      "loss": 1.5956,
      "step": 34
    },
    {
      "epoch": 0.013983220135836995,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005,
      "loss": 1.5822,
      "step": 35
    },
    {
      "epoch": 0.014382740711146624,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005,
      "loss": 1.5709,
      "step": 36
    },
    {
      "epoch": 0.014782261286456252,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0005,
      "loss": 1.5436,
      "step": 37
    },
    {
      "epoch": 0.015181781861765881,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005,
      "loss": 1.5349,
      "step": 38
    },
    {
      "epoch": 0.01558130243707551,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005,
      "loss": 1.5121,
      "step": 39
    },
    {
      "epoch": 0.015980823012385136,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0005,
      "loss": 1.5008,
      "step": 40
    },
    {
      "epoch": 0.016380343587694766,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0005,
      "loss": 1.5031,
      "step": 41
    },
    {
      "epoch": 0.016779864163004393,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0005,
      "loss": 1.4996,
      "step": 42
    },
    {
      "epoch": 0.017179384738314023,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005,
      "loss": 1.4847,
      "step": 43
    },
    {
      "epoch": 0.01757890531362365,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005,
      "loss": 1.4572,
      "step": 44
    },
    {
      "epoch": 0.01797842588893328,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005,
      "loss": 1.4555,
      "step": 45
    },
    {
      "epoch": 0.018377946464242907,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005,
      "loss": 1.4598,
      "step": 46
    },
    {
      "epoch": 0.018777467039552537,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005,
      "loss": 1.4292,
      "step": 47
    },
    {
      "epoch": 0.019176987614862164,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005,
      "loss": 1.4271,
      "step": 48
    },
    {
      "epoch": 0.019576508190171794,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005,
      "loss": 1.4139,
      "step": 49
    },
    {
      "epoch": 0.01997602876548142,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0005,
      "loss": 1.4094,
      "step": 50
    },
    {
      "epoch": 0.02037554934079105,
      "grad_norm": 0.375,
      "learning_rate": 0.0005,
      "loss": 1.4106,
      "step": 51
    },
    {
      "epoch": 0.020775069916100678,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0005,
      "loss": 1.399,
      "step": 52
    },
    {
      "epoch": 0.02117459049141031,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005,
      "loss": 1.3949,
      "step": 53
    },
    {
      "epoch": 0.021574111066719935,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0005,
      "loss": 1.3915,
      "step": 54
    },
    {
      "epoch": 0.021973631642029565,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0005,
      "loss": 1.3763,
      "step": 55
    },
    {
      "epoch": 0.022373152217339192,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005,
      "loss": 1.3636,
      "step": 56
    },
    {
      "epoch": 0.022772672792648822,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005,
      "loss": 1.3694,
      "step": 57
    },
    {
      "epoch": 0.02317219336795845,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0005,
      "loss": 1.3668,
      "step": 58
    },
    {
      "epoch": 0.02357171394326808,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0005,
      "loss": 1.3664,
      "step": 59
    },
    {
      "epoch": 0.023971234518577706,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0005,
      "loss": 1.3588,
      "step": 60
    },
    {
      "epoch": 0.024370755093887336,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0005,
      "loss": 1.356,
      "step": 61
    },
    {
      "epoch": 0.024770275669196963,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0005,
      "loss": 1.3621,
      "step": 62
    },
    {
      "epoch": 0.025169796244506593,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0005,
      "loss": 1.3493,
      "step": 63
    },
    {
      "epoch": 0.02556931681981622,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005,
      "loss": 1.3488,
      "step": 64
    },
    {
      "epoch": 0.02596883739512585,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0005,
      "loss": 1.3591,
      "step": 65
    },
    {
      "epoch": 0.026368357970435477,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0005,
      "loss": 1.3558,
      "step": 66
    },
    {
      "epoch": 0.026767878545745107,
      "grad_norm": 0.6875,
      "learning_rate": 0.0005,
      "loss": 1.3593,
      "step": 67
    },
    {
      "epoch": 0.027167399121054734,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005,
      "loss": 1.3227,
      "step": 68
    },
    {
      "epoch": 0.027566919696364364,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0005,
      "loss": 1.3502,
      "step": 69
    },
    {
      "epoch": 0.02796644027167399,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0005,
      "loss": 1.3437,
      "step": 70
    },
    {
      "epoch": 0.02836596084698362,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0005,
      "loss": 1.3553,
      "step": 71
    },
    {
      "epoch": 0.028765481422293248,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005,
      "loss": 1.3279,
      "step": 72
    },
    {
      "epoch": 0.029165001997602878,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005,
      "loss": 1.3416,
      "step": 73
    },
    {
      "epoch": 0.029564522572912505,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005,
      "loss": 1.3436,
      "step": 74
    },
    {
      "epoch": 0.02996404314822213,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005,
      "loss": 1.3348,
      "step": 75
    },
    {
      "epoch": 0.030363563723531762,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0005,
      "loss": 1.3171,
      "step": 76
    },
    {
      "epoch": 0.03076308429884139,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0005,
      "loss": 1.3143,
      "step": 77
    },
    {
      "epoch": 0.03116260487415102,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0005,
      "loss": 1.3066,
      "step": 78
    },
    {
      "epoch": 0.031562125449460646,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0005,
      "loss": 1.3274,
      "step": 79
    },
    {
      "epoch": 0.03196164602477027,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0005,
      "loss": 1.3102,
      "step": 80
    },
    {
      "epoch": 0.032361166600079906,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0005,
      "loss": 1.312,
      "step": 81
    },
    {
      "epoch": 0.03276068717538953,
      "grad_norm": 0.28125,
      "learning_rate": 0.0005,
      "loss": 1.3134,
      "step": 82
    },
    {
      "epoch": 0.03316020775069916,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0005,
      "loss": 1.2988,
      "step": 83
    },
    {
      "epoch": 0.033559728326008786,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0005,
      "loss": 1.3042,
      "step": 84
    },
    {
      "epoch": 0.03395924890131842,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0005,
      "loss": 1.3062,
      "step": 85
    },
    {
      "epoch": 0.03435876947662805,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0005,
      "loss": 1.2982,
      "step": 86
    },
    {
      "epoch": 0.034758290051937674,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0005,
      "loss": 1.3077,
      "step": 87
    },
    {
      "epoch": 0.0351578106272473,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005,
      "loss": 1.3042,
      "step": 88
    },
    {
      "epoch": 0.035557331202556934,
      "grad_norm": 0.40625,
      "learning_rate": 0.0005,
      "loss": 1.2904,
      "step": 89
    },
    {
      "epoch": 0.03595685177786656,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0005,
      "loss": 1.2845,
      "step": 90
    },
    {
      "epoch": 0.03635637235317619,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0005,
      "loss": 1.3028,
      "step": 91
    },
    {
      "epoch": 0.036755892928485814,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0005,
      "loss": 1.2846,
      "step": 92
    },
    {
      "epoch": 0.03715541350379545,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0005,
      "loss": 1.281,
      "step": 93
    },
    {
      "epoch": 0.037554934079105075,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0005,
      "loss": 1.2841,
      "step": 94
    },
    {
      "epoch": 0.0379544546544147,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0005,
      "loss": 1.2928,
      "step": 95
    },
    {
      "epoch": 0.03835397522972433,
      "grad_norm": 0.265625,
      "learning_rate": 0.0005,
      "loss": 1.2807,
      "step": 96
    },
    {
      "epoch": 0.03875349580503396,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0005,
      "loss": 1.2837,
      "step": 97
    },
    {
      "epoch": 0.03915301638034359,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0005,
      "loss": 1.2684,
      "step": 98
    },
    {
      "epoch": 0.039552536955653216,
      "grad_norm": 0.234375,
      "learning_rate": 0.0005,
      "loss": 1.2694,
      "step": 99
    },
    {
      "epoch": 0.03995205753096284,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0005,
      "loss": 1.2613,
      "step": 100
    },
    {
      "epoch": 0.040351578106272476,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.2594,
      "step": 101
    },
    {
      "epoch": 0.0407510986815821,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0005,
      "loss": 1.2697,
      "step": 102
    },
    {
      "epoch": 0.04115061925689173,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0005,
      "loss": 1.2636,
      "step": 103
    },
    {
      "epoch": 0.041550139832201356,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0005,
      "loss": 1.2633,
      "step": 104
    },
    {
      "epoch": 0.04194966040751099,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005,
      "loss": 1.2651,
      "step": 105
    },
    {
      "epoch": 0.04234918098282062,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005,
      "loss": 1.2696,
      "step": 106
    },
    {
      "epoch": 0.042748701558130243,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0005,
      "loss": 1.2726,
      "step": 107
    },
    {
      "epoch": 0.04314822213343987,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0005,
      "loss": 1.2519,
      "step": 108
    },
    {
      "epoch": 0.043547742708749504,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0005,
      "loss": 1.2695,
      "step": 109
    },
    {
      "epoch": 0.04394726328405913,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005,
      "loss": 1.2667,
      "step": 110
    },
    {
      "epoch": 0.04434678385936876,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005,
      "loss": 1.2507,
      "step": 111
    },
    {
      "epoch": 0.044746304434678384,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0005,
      "loss": 1.2676,
      "step": 112
    },
    {
      "epoch": 0.04514582500998801,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005,
      "loss": 1.2574,
      "step": 113
    },
    {
      "epoch": 0.045545345585297645,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0005,
      "loss": 1.2484,
      "step": 114
    },
    {
      "epoch": 0.04594486616060727,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005,
      "loss": 1.2514,
      "step": 115
    },
    {
      "epoch": 0.0463443867359169,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0005,
      "loss": 1.2374,
      "step": 116
    },
    {
      "epoch": 0.046743907311226525,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0005,
      "loss": 1.2643,
      "step": 117
    },
    {
      "epoch": 0.04714342788653616,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0005,
      "loss": 1.2525,
      "step": 118
    },
    {
      "epoch": 0.047542948461845785,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0005,
      "loss": 1.2457,
      "step": 119
    },
    {
      "epoch": 0.04794246903715541,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0005,
      "loss": 1.2422,
      "step": 120
    },
    {
      "epoch": 0.04834198961246504,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0005,
      "loss": 1.2451,
      "step": 121
    },
    {
      "epoch": 0.04874151018777467,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0005,
      "loss": 1.2453,
      "step": 122
    },
    {
      "epoch": 0.0491410307630843,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0005,
      "loss": 1.2419,
      "step": 123
    },
    {
      "epoch": 0.049540551338393926,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0005,
      "loss": 1.2361,
      "step": 124
    },
    {
      "epoch": 0.04994007191370355,
      "grad_norm": 0.28125,
      "learning_rate": 0.0005,
      "loss": 1.221,
      "step": 125
    },
    {
      "epoch": 0.05033959248901319,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0005,
      "loss": 1.238,
      "step": 126
    },
    {
      "epoch": 0.05073911306432281,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.2425,
      "step": 127
    },
    {
      "epoch": 0.05113863363963244,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0005,
      "loss": 1.2299,
      "step": 128
    },
    {
      "epoch": 0.05153815421494207,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0005,
      "loss": 1.2447,
      "step": 129
    },
    {
      "epoch": 0.0519376747902517,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0005,
      "loss": 1.2401,
      "step": 130
    },
    {
      "epoch": 0.05233719536556133,
      "grad_norm": 0.3125,
      "learning_rate": 0.0005,
      "loss": 1.2332,
      "step": 131
    },
    {
      "epoch": 0.052736715940870954,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0005,
      "loss": 1.2308,
      "step": 132
    },
    {
      "epoch": 0.05313623651618058,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005,
      "loss": 1.2323,
      "step": 133
    },
    {
      "epoch": 0.053535757091490214,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005,
      "loss": 1.2371,
      "step": 134
    },
    {
      "epoch": 0.05393527766679984,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005,
      "loss": 1.2282,
      "step": 135
    },
    {
      "epoch": 0.05433479824210947,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0005,
      "loss": 1.2391,
      "step": 136
    },
    {
      "epoch": 0.054734318817419095,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005,
      "loss": 1.2467,
      "step": 137
    },
    {
      "epoch": 0.05513383939272873,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005,
      "loss": 1.2266,
      "step": 138
    },
    {
      "epoch": 0.055533359968038355,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0005,
      "loss": 1.2468,
      "step": 139
    },
    {
      "epoch": 0.05593288054334798,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0005,
      "loss": 1.2299,
      "step": 140
    },
    {
      "epoch": 0.05633240111865761,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0005,
      "loss": 1.2147,
      "step": 141
    },
    {
      "epoch": 0.05673192169396724,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0005,
      "loss": 1.234,
      "step": 142
    },
    {
      "epoch": 0.05713144226927687,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0005,
      "loss": 1.2296,
      "step": 143
    },
    {
      "epoch": 0.057530962844586496,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0005,
      "loss": 1.2334,
      "step": 144
    },
    {
      "epoch": 0.05793048341989612,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0005,
      "loss": 1.2367,
      "step": 145
    },
    {
      "epoch": 0.058330003995205756,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0005,
      "loss": 1.2225,
      "step": 146
    },
    {
      "epoch": 0.05872952457051538,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0005,
      "loss": 1.2325,
      "step": 147
    },
    {
      "epoch": 0.05912904514582501,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0005,
      "loss": 1.2237,
      "step": 148
    },
    {
      "epoch": 0.05952856572113464,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0005,
      "loss": 1.2226,
      "step": 149
    },
    {
      "epoch": 0.05992808629644426,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0005,
      "loss": 1.225,
      "step": 150
    },
    {
      "epoch": 0.0603276068717539,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0005,
      "loss": 1.2315,
      "step": 151
    },
    {
      "epoch": 0.060727127447063524,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0005,
      "loss": 1.2224,
      "step": 152
    },
    {
      "epoch": 0.06112664802237315,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0005,
      "loss": 1.2152,
      "step": 153
    },
    {
      "epoch": 0.06152616859768278,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0005,
      "loss": 1.2128,
      "step": 154
    },
    {
      "epoch": 0.06192568917299241,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0005,
      "loss": 1.2218,
      "step": 155
    },
    {
      "epoch": 0.06232520974830204,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005,
      "loss": 1.2318,
      "step": 156
    },
    {
      "epoch": 0.06272473032361167,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0005,
      "loss": 1.2094,
      "step": 157
    },
    {
      "epoch": 0.06312425089892129,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005,
      "loss": 1.2367,
      "step": 158
    },
    {
      "epoch": 0.06352377147423093,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0005,
      "loss": 1.2254,
      "step": 159
    },
    {
      "epoch": 0.06392329204954054,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0005,
      "loss": 1.23,
      "step": 160
    },
    {
      "epoch": 0.06432281262485018,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0005,
      "loss": 1.2304,
      "step": 161
    },
    {
      "epoch": 0.06472233320015981,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0005,
      "loss": 1.2216,
      "step": 162
    },
    {
      "epoch": 0.06512185377546943,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0005,
      "loss": 1.2211,
      "step": 163
    },
    {
      "epoch": 0.06552137435077907,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0005,
      "loss": 1.2095,
      "step": 164
    },
    {
      "epoch": 0.0659208949260887,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0005,
      "loss": 1.2175,
      "step": 165
    },
    {
      "epoch": 0.06632041550139832,
      "grad_norm": 0.296875,
      "learning_rate": 0.0005,
      "loss": 1.201,
      "step": 166
    },
    {
      "epoch": 0.06671993607670795,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0005,
      "loss": 1.214,
      "step": 167
    },
    {
      "epoch": 0.06711945665201757,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0005,
      "loss": 1.2096,
      "step": 168
    },
    {
      "epoch": 0.0675189772273272,
      "grad_norm": 0.25,
      "learning_rate": 0.0005,
      "loss": 1.2102,
      "step": 169
    },
    {
      "epoch": 0.06791849780263684,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.2061,
      "step": 170
    },
    {
      "epoch": 0.06831801837794646,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0005,
      "loss": 1.2017,
      "step": 171
    },
    {
      "epoch": 0.0687175389532561,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0005,
      "loss": 1.1911,
      "step": 172
    },
    {
      "epoch": 0.06911705952856573,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0005,
      "loss": 1.2196,
      "step": 173
    },
    {
      "epoch": 0.06951658010387535,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0005,
      "loss": 1.2002,
      "step": 174
    },
    {
      "epoch": 0.06991610067918498,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0005,
      "loss": 1.2146,
      "step": 175
    },
    {
      "epoch": 0.0703156212544946,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005,
      "loss": 1.217,
      "step": 176
    },
    {
      "epoch": 0.07071514182980423,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0005,
      "loss": 1.1998,
      "step": 177
    },
    {
      "epoch": 0.07111466240511387,
      "grad_norm": 0.25,
      "learning_rate": 0.0005,
      "loss": 1.2189,
      "step": 178
    },
    {
      "epoch": 0.07151418298042349,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0005,
      "loss": 1.2081,
      "step": 179
    },
    {
      "epoch": 0.07191370355573312,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0005,
      "loss": 1.2105,
      "step": 180
    },
    {
      "epoch": 0.07231322413104276,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0005,
      "loss": 1.2148,
      "step": 181
    },
    {
      "epoch": 0.07271274470635238,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0005,
      "loss": 1.1977,
      "step": 182
    },
    {
      "epoch": 0.07311226528166201,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0005,
      "loss": 1.2027,
      "step": 183
    },
    {
      "epoch": 0.07351178585697163,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0005,
      "loss": 1.2038,
      "step": 184
    },
    {
      "epoch": 0.07391130643228126,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0005,
      "loss": 1.207,
      "step": 185
    },
    {
      "epoch": 0.0743108270075909,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0005,
      "loss": 1.2025,
      "step": 186
    },
    {
      "epoch": 0.07471034758290052,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0005,
      "loss": 1.1999,
      "step": 187
    },
    {
      "epoch": 0.07510986815821015,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0005,
      "loss": 1.2108,
      "step": 188
    },
    {
      "epoch": 0.07550938873351977,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0005,
      "loss": 1.2082,
      "step": 189
    },
    {
      "epoch": 0.0759089093088294,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0005,
      "loss": 1.2076,
      "step": 190
    },
    {
      "epoch": 0.07630842988413904,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0005,
      "loss": 1.2073,
      "step": 191
    },
    {
      "epoch": 0.07670795045944866,
      "grad_norm": 0.265625,
      "learning_rate": 0.0005,
      "loss": 1.1904,
      "step": 192
    },
    {
      "epoch": 0.07710747103475829,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0005,
      "loss": 1.2029,
      "step": 193
    },
    {
      "epoch": 0.07750699161006792,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0005,
      "loss": 1.1945,
      "step": 194
    },
    {
      "epoch": 0.07790651218537754,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0005,
      "loss": 1.2095,
      "step": 195
    },
    {
      "epoch": 0.07830603276068718,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0005,
      "loss": 1.2029,
      "step": 196
    },
    {
      "epoch": 0.0787055533359968,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0005,
      "loss": 1.2025,
      "step": 197
    },
    {
      "epoch": 0.07910507391130643,
      "grad_norm": 0.265625,
      "learning_rate": 0.0005,
      "loss": 1.1891,
      "step": 198
    },
    {
      "epoch": 0.07950459448661606,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.1928,
      "step": 199
    },
    {
      "epoch": 0.07990411506192568,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0005,
      "loss": 1.2055,
      "step": 200
    },
    {
      "epoch": 0.08030363563723532,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0005,
      "loss": 1.2098,
      "step": 201
    },
    {
      "epoch": 0.08070315621254495,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0005,
      "loss": 1.1928,
      "step": 202
    },
    {
      "epoch": 0.08110267678785457,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0005,
      "loss": 1.1964,
      "step": 203
    },
    {
      "epoch": 0.0815021973631642,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0005,
      "loss": 1.1919,
      "step": 204
    },
    {
      "epoch": 0.08190171793847383,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0005,
      "loss": 1.1907,
      "step": 205
    },
    {
      "epoch": 0.08230123851378346,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0005,
      "loss": 1.1992,
      "step": 206
    },
    {
      "epoch": 0.08270075908909309,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0005,
      "loss": 1.2075,
      "step": 207
    },
    {
      "epoch": 0.08310027966440271,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0005,
      "loss": 1.2005,
      "step": 208
    },
    {
      "epoch": 0.08349980023971235,
      "grad_norm": 0.28125,
      "learning_rate": 0.0005,
      "loss": 1.1882,
      "step": 209
    },
    {
      "epoch": 0.08389932081502198,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0005,
      "loss": 1.1984,
      "step": 210
    },
    {
      "epoch": 0.0842988413903316,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0005,
      "loss": 1.2138,
      "step": 211
    },
    {
      "epoch": 0.08469836196564123,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0005,
      "loss": 1.1859,
      "step": 212
    },
    {
      "epoch": 0.08509788254095085,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0005,
      "loss": 1.1894,
      "step": 213
    },
    {
      "epoch": 0.08549740311626049,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005,
      "loss": 1.1867,
      "step": 214
    },
    {
      "epoch": 0.08589692369157012,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005,
      "loss": 1.2184,
      "step": 215
    },
    {
      "epoch": 0.08629644426687974,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005,
      "loss": 1.2526,
      "step": 216
    },
    {
      "epoch": 0.08669596484218937,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0005,
      "loss": 1.2766,
      "step": 217
    },
    {
      "epoch": 0.08709548541749901,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0005,
      "loss": 1.2506,
      "step": 218
    },
    {
      "epoch": 0.08749500599280863,
      "grad_norm": 0.578125,
      "learning_rate": 0.0005,
      "loss": 1.255,
      "step": 219
    },
    {
      "epoch": 0.08789452656811826,
      "grad_norm": 0.3125,
      "learning_rate": 0.0005,
      "loss": 1.2583,
      "step": 220
    },
    {
      "epoch": 0.08829404714342788,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0005,
      "loss": 1.2486,
      "step": 221
    },
    {
      "epoch": 0.08869356771873751,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005,
      "loss": 1.2367,
      "step": 222
    },
    {
      "epoch": 0.08909308829404715,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0005,
      "loss": 1.2234,
      "step": 223
    },
    {
      "epoch": 0.08949260886935677,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005,
      "loss": 1.2375,
      "step": 224
    },
    {
      "epoch": 0.0898921294446664,
      "grad_norm": 0.234375,
      "learning_rate": 0.0005,
      "loss": 1.236,
      "step": 225
    },
    {
      "epoch": 0.09029165001997602,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0005,
      "loss": 1.2155,
      "step": 226
    },
    {
      "epoch": 0.09069117059528566,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0005,
      "loss": 1.2178,
      "step": 227
    },
    {
      "epoch": 0.09109069117059529,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0005,
      "loss": 1.2117,
      "step": 228
    },
    {
      "epoch": 0.09149021174590491,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0005,
      "loss": 1.2235,
      "step": 229
    },
    {
      "epoch": 0.09188973232121454,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0005,
      "loss": 1.2189,
      "step": 230
    },
    {
      "epoch": 0.09228925289652418,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0005,
      "loss": 1.2079,
      "step": 231
    },
    {
      "epoch": 0.0926887734718338,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0005,
      "loss": 1.2094,
      "step": 232
    },
    {
      "epoch": 0.09308829404714343,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005,
      "loss": 1.1981,
      "step": 233
    },
    {
      "epoch": 0.09348781462245305,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0005,
      "loss": 1.2178,
      "step": 234
    },
    {
      "epoch": 0.09388733519776268,
      "grad_norm": 0.21875,
      "learning_rate": 0.0005,
      "loss": 1.2071,
      "step": 235
    },
    {
      "epoch": 0.09428685577307232,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0005,
      "loss": 1.1894,
      "step": 236
    },
    {
      "epoch": 0.09468637634838194,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0005,
      "loss": 1.192,
      "step": 237
    },
    {
      "epoch": 0.09508589692369157,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0005,
      "loss": 1.199,
      "step": 238
    },
    {
      "epoch": 0.0954854174990012,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0005,
      "loss": 1.2165,
      "step": 239
    },
    {
      "epoch": 0.09588493807431082,
      "grad_norm": 0.21875,
      "learning_rate": 0.0005,
      "loss": 1.1975,
      "step": 240
    },
    {
      "epoch": 0.09628445864962046,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0005,
      "loss": 1.1948,
      "step": 241
    },
    {
      "epoch": 0.09668397922493008,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0005,
      "loss": 1.1932,
      "step": 242
    },
    {
      "epoch": 0.09708349980023971,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0005,
      "loss": 1.1978,
      "step": 243
    },
    {
      "epoch": 0.09748302037554935,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0005,
      "loss": 1.1932,
      "step": 244
    },
    {
      "epoch": 0.09788254095085897,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0005,
      "loss": 1.1988,
      "step": 245
    },
    {
      "epoch": 0.0982820615261686,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0005,
      "loss": 1.2009,
      "step": 246
    },
    {
      "epoch": 0.09868158210147823,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0005,
      "loss": 1.2003,
      "step": 247
    },
    {
      "epoch": 0.09908110267678785,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0005,
      "loss": 1.2001,
      "step": 248
    },
    {
      "epoch": 0.09948062325209749,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0005,
      "loss": 1.19,
      "step": 249
    },
    {
      "epoch": 0.0998801438274071,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0005,
      "loss": 1.1888,
      "step": 250
    },
    {
      "epoch": 0.10027966440271674,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0005,
      "loss": 1.1851,
      "step": 251
    },
    {
      "epoch": 0.10067918497802637,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.179,
      "step": 252
    },
    {
      "epoch": 0.10107870555333599,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0005,
      "loss": 1.1999,
      "step": 253
    },
    {
      "epoch": 0.10147822612864563,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005,
      "loss": 1.2023,
      "step": 254
    },
    {
      "epoch": 0.10187774670395526,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0005,
      "loss": 1.1998,
      "step": 255
    },
    {
      "epoch": 0.10227726727926488,
      "grad_norm": 0.28125,
      "learning_rate": 0.0005,
      "loss": 1.1926,
      "step": 256
    },
    {
      "epoch": 0.10267678785457451,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0005,
      "loss": 1.1924,
      "step": 257
    },
    {
      "epoch": 0.10307630842988413,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0005,
      "loss": 1.1782,
      "step": 258
    },
    {
      "epoch": 0.10347582900519377,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.198,
      "step": 259
    },
    {
      "epoch": 0.1038753495805034,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0005,
      "loss": 1.1839,
      "step": 260
    },
    {
      "epoch": 0.10427487015581302,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0005,
      "loss": 1.1991,
      "step": 261
    },
    {
      "epoch": 0.10467439073112265,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0005,
      "loss": 1.185,
      "step": 262
    },
    {
      "epoch": 0.10507391130643227,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0005,
      "loss": 1.1815,
      "step": 263
    },
    {
      "epoch": 0.10547343188174191,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0005,
      "loss": 1.1967,
      "step": 264
    },
    {
      "epoch": 0.10587295245705154,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0005,
      "loss": 1.1831,
      "step": 265
    },
    {
      "epoch": 0.10627247303236116,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0005,
      "loss": 1.1872,
      "step": 266
    },
    {
      "epoch": 0.1066719936076708,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0005,
      "loss": 1.1988,
      "step": 267
    },
    {
      "epoch": 0.10707151418298043,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0005,
      "loss": 1.1834,
      "step": 268
    },
    {
      "epoch": 0.10747103475829005,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.1782,
      "step": 269
    },
    {
      "epoch": 0.10787055533359968,
      "grad_norm": 0.265625,
      "learning_rate": 0.0005,
      "loss": 1.1779,
      "step": 270
    },
    {
      "epoch": 0.1082700759089093,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0005,
      "loss": 1.1743,
      "step": 271
    },
    {
      "epoch": 0.10866959648421894,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0005,
      "loss": 1.1861,
      "step": 272
    },
    {
      "epoch": 0.10906911705952857,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0005,
      "loss": 1.1875,
      "step": 273
    },
    {
      "epoch": 0.10946863763483819,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0005,
      "loss": 1.1697,
      "step": 274
    },
    {
      "epoch": 0.10986815821014782,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0005,
      "loss": 1.1796,
      "step": 275
    },
    {
      "epoch": 0.11026767878545746,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0005,
      "loss": 1.1808,
      "step": 276
    },
    {
      "epoch": 0.11066719936076708,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0005,
      "loss": 1.2016,
      "step": 277
    },
    {
      "epoch": 0.11106671993607671,
      "grad_norm": 0.796875,
      "learning_rate": 0.0005,
      "loss": 1.2221,
      "step": 278
    },
    {
      "epoch": 0.11146624051138633,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0005,
      "loss": 1.2108,
      "step": 279
    },
    {
      "epoch": 0.11186576108669596,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0005,
      "loss": 1.1984,
      "step": 280
    },
    {
      "epoch": 0.1122652816620056,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.2073,
      "step": 281
    },
    {
      "epoch": 0.11266480223731522,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005,
      "loss": 1.1945,
      "step": 282
    },
    {
      "epoch": 0.11306432281262485,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0005,
      "loss": 1.184,
      "step": 283
    },
    {
      "epoch": 0.11346384338793448,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0005,
      "loss": 1.1903,
      "step": 284
    },
    {
      "epoch": 0.1138633639632441,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0005,
      "loss": 1.1902,
      "step": 285
    },
    {
      "epoch": 0.11426288453855374,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0005,
      "loss": 1.1941,
      "step": 286
    },
    {
      "epoch": 0.11466240511386336,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0005,
      "loss": 1.2027,
      "step": 287
    },
    {
      "epoch": 0.11506192568917299,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0005,
      "loss": 1.1756,
      "step": 288
    },
    {
      "epoch": 0.11546144626448263,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.1766,
      "step": 289
    },
    {
      "epoch": 0.11586096683979225,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.1671,
      "step": 290
    },
    {
      "epoch": 0.11626048741510188,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.1756,
      "step": 291
    },
    {
      "epoch": 0.11666000799041151,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.2036,
      "step": 292
    },
    {
      "epoch": 0.11705952856572113,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.1797,
      "step": 293
    },
    {
      "epoch": 0.11745904914103077,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.1793,
      "step": 294
    },
    {
      "epoch": 0.11785856971634039,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.177,
      "step": 295
    },
    {
      "epoch": 0.11825809029165002,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0005,
      "loss": 1.178,
      "step": 296
    },
    {
      "epoch": 0.11865761086695965,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.1697,
      "step": 297
    },
    {
      "epoch": 0.11905713144226927,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.1791,
      "step": 298
    },
    {
      "epoch": 0.11945665201757891,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.1904,
      "step": 299
    },
    {
      "epoch": 0.11985617259288853,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.1701,
      "step": 300
    },
    {
      "epoch": 0.12025569316819816,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.1809,
      "step": 301
    },
    {
      "epoch": 0.1206552137435078,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0005,
      "loss": 1.1686,
      "step": 302
    },
    {
      "epoch": 0.12105473431881741,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.1651,
      "step": 303
    },
    {
      "epoch": 0.12145425489412705,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0005,
      "loss": 1.1712,
      "step": 304
    },
    {
      "epoch": 0.12185377546943668,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.1784,
      "step": 305
    },
    {
      "epoch": 0.1222532960447463,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.1764,
      "step": 306
    },
    {
      "epoch": 0.12265281662005593,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.1511,
      "step": 307
    },
    {
      "epoch": 0.12305233719536555,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.1663,
      "step": 308
    },
    {
      "epoch": 0.12345185777067519,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.1722,
      "step": 309
    },
    {
      "epoch": 0.12385137834598482,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0005,
      "loss": 1.1696,
      "step": 310
    },
    {
      "epoch": 0.12425089892129444,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.164,
      "step": 311
    },
    {
      "epoch": 0.12465041949660408,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0005,
      "loss": 1.1665,
      "step": 312
    },
    {
      "epoch": 0.1250499400719137,
      "grad_norm": 0.265625,
      "learning_rate": 0.0005,
      "loss": 1.1808,
      "step": 313
    },
    {
      "epoch": 0.12544946064722334,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0005,
      "loss": 1.1648,
      "step": 314
    },
    {
      "epoch": 0.12584898122253296,
      "grad_norm": 0.28125,
      "learning_rate": 0.0005,
      "loss": 1.1614,
      "step": 315
    },
    {
      "epoch": 0.12624850179784258,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0005,
      "loss": 1.1653,
      "step": 316
    },
    {
      "epoch": 0.12664802237315223,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0005,
      "loss": 1.1657,
      "step": 317
    },
    {
      "epoch": 0.12704754294846185,
      "grad_norm": 0.265625,
      "learning_rate": 0.0005,
      "loss": 1.1511,
      "step": 318
    },
    {
      "epoch": 0.12744706352377147,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0005,
      "loss": 1.1654,
      "step": 319
    },
    {
      "epoch": 0.1278465840990811,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0005,
      "loss": 1.1677,
      "step": 320
    },
    {
      "epoch": 0.12824610467439074,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005,
      "loss": 1.1568,
      "step": 321
    },
    {
      "epoch": 0.12864562524970036,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0005,
      "loss": 1.1775,
      "step": 322
    },
    {
      "epoch": 0.12904514582500998,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0005,
      "loss": 1.186,
      "step": 323
    },
    {
      "epoch": 0.12944466640031962,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0005,
      "loss": 1.1871,
      "step": 324
    },
    {
      "epoch": 0.12984418697562924,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0005,
      "loss": 1.1934,
      "step": 325
    },
    {
      "epoch": 0.13024370755093886,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0005,
      "loss": 1.1796,
      "step": 326
    },
    {
      "epoch": 0.1306432281262485,
      "grad_norm": 0.28125,
      "learning_rate": 0.0005,
      "loss": 1.1741,
      "step": 327
    },
    {
      "epoch": 0.13104274870155813,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.1744,
      "step": 328
    },
    {
      "epoch": 0.13144226927686775,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0005,
      "loss": 1.1802,
      "step": 329
    },
    {
      "epoch": 0.1318417898521774,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005,
      "loss": 1.172,
      "step": 330
    },
    {
      "epoch": 0.13224131042748702,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.1667,
      "step": 331
    },
    {
      "epoch": 0.13264083100279664,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.1635,
      "step": 332
    },
    {
      "epoch": 0.1330403515781063,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.1617,
      "step": 333
    },
    {
      "epoch": 0.1334398721534159,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0005,
      "loss": 1.1657,
      "step": 334
    },
    {
      "epoch": 0.13383939272872553,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.1628,
      "step": 335
    },
    {
      "epoch": 0.13423891330403515,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.17,
      "step": 336
    },
    {
      "epoch": 0.1346384338793448,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0005,
      "loss": 1.1576,
      "step": 337
    },
    {
      "epoch": 0.1350379544546544,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0005,
      "loss": 1.17,
      "step": 338
    },
    {
      "epoch": 0.13543747502996403,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0005,
      "loss": 1.1678,
      "step": 339
    },
    {
      "epoch": 0.13583699560527368,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0005,
      "loss": 1.1596,
      "step": 340
    },
    {
      "epoch": 0.1362365161805833,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0005,
      "loss": 1.1674,
      "step": 341
    },
    {
      "epoch": 0.13663603675589292,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0005,
      "loss": 1.1605,
      "step": 342
    },
    {
      "epoch": 0.13703555733120257,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0005,
      "loss": 1.1593,
      "step": 343
    },
    {
      "epoch": 0.1374350779065122,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0005,
      "loss": 1.1692,
      "step": 344
    },
    {
      "epoch": 0.1378345984818218,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0005,
      "loss": 1.1635,
      "step": 345
    },
    {
      "epoch": 0.13823411905713145,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.1509,
      "step": 346
    },
    {
      "epoch": 0.13863363963244107,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0005,
      "loss": 1.1606,
      "step": 347
    },
    {
      "epoch": 0.1390331602077507,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.1482,
      "step": 348
    },
    {
      "epoch": 0.13943268078306031,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0005,
      "loss": 1.1624,
      "step": 349
    },
    {
      "epoch": 0.13983220135836996,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.1525,
      "step": 350
    },
    {
      "epoch": 0.14023172193367958,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0005,
      "loss": 1.1568,
      "step": 351
    },
    {
      "epoch": 0.1406312425089892,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005,
      "loss": 1.1618,
      "step": 352
    },
    {
      "epoch": 0.14103076308429885,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.1506,
      "step": 353
    },
    {
      "epoch": 0.14143028365960847,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0005,
      "loss": 1.1542,
      "step": 354
    },
    {
      "epoch": 0.1418298042349181,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.1482,
      "step": 355
    },
    {
      "epoch": 0.14222932481022774,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.1621,
      "step": 356
    },
    {
      "epoch": 0.14262884538553736,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.1473,
      "step": 357
    },
    {
      "epoch": 0.14302836596084698,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.1633,
      "step": 358
    },
    {
      "epoch": 0.14342788653615662,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.1494,
      "step": 359
    },
    {
      "epoch": 0.14382740711146624,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.1562,
      "step": 360
    },
    {
      "epoch": 0.14422692768677586,
      "grad_norm": 0.203125,
      "learning_rate": 0.0005,
      "loss": 1.156,
      "step": 361
    },
    {
      "epoch": 0.1446264482620855,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.1581,
      "step": 362
    },
    {
      "epoch": 0.14502596883739513,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.1488,
      "step": 363
    },
    {
      "epoch": 0.14542548941270475,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.1453,
      "step": 364
    },
    {
      "epoch": 0.14582500998801437,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.1601,
      "step": 365
    },
    {
      "epoch": 0.14622453056332402,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.153,
      "step": 366
    },
    {
      "epoch": 0.14662405113863364,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.1548,
      "step": 367
    },
    {
      "epoch": 0.14702357171394326,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.1584,
      "step": 368
    },
    {
      "epoch": 0.1474230922892529,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.1517,
      "step": 369
    },
    {
      "epoch": 0.14782261286456252,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.1503,
      "step": 370
    },
    {
      "epoch": 0.14822213343987214,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.1494,
      "step": 371
    },
    {
      "epoch": 0.1486216540151818,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.1514,
      "step": 372
    },
    {
      "epoch": 0.1490211745904914,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.1394,
      "step": 373
    },
    {
      "epoch": 0.14942069516580103,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0005,
      "loss": 1.1533,
      "step": 374
    },
    {
      "epoch": 0.14982021574111068,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.1571,
      "step": 375
    },
    {
      "epoch": 0.1502197363164203,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0005,
      "loss": 1.1485,
      "step": 376
    },
    {
      "epoch": 0.15061925689172992,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0005,
      "loss": 1.1492,
      "step": 377
    },
    {
      "epoch": 0.15101877746703954,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.1444,
      "step": 378
    },
    {
      "epoch": 0.1514182980423492,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.1514,
      "step": 379
    },
    {
      "epoch": 0.1518178186176588,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.1573,
      "step": 380
    },
    {
      "epoch": 0.15221733919296843,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.1579,
      "step": 381
    },
    {
      "epoch": 0.15261685976827807,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.156,
      "step": 382
    },
    {
      "epoch": 0.1530163803435877,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.1511,
      "step": 383
    },
    {
      "epoch": 0.1534159009188973,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.1443,
      "step": 384
    },
    {
      "epoch": 0.15381542149420696,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0005,
      "loss": 1.1468,
      "step": 385
    },
    {
      "epoch": 0.15421494206951658,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0005,
      "loss": 1.144,
      "step": 386
    },
    {
      "epoch": 0.1546144626448262,
      "grad_norm": 1.40625,
      "learning_rate": 0.0005,
      "loss": 1.1898,
      "step": 387
    },
    {
      "epoch": 0.15501398322013585,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005,
      "loss": 1.2046,
      "step": 388
    },
    {
      "epoch": 0.15541350379544547,
      "grad_norm": 0.4375,
      "learning_rate": 0.0005,
      "loss": 1.1693,
      "step": 389
    },
    {
      "epoch": 0.1558130243707551,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0005,
      "loss": 1.1749,
      "step": 390
    },
    {
      "epoch": 0.15621254494606474,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0005,
      "loss": 1.177,
      "step": 391
    },
    {
      "epoch": 0.15661206552137436,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0005,
      "loss": 1.1668,
      "step": 392
    },
    {
      "epoch": 0.15701158609668397,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0005,
      "loss": 1.1896,
      "step": 393
    },
    {
      "epoch": 0.1574111066719936,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0005,
      "loss": 1.1686,
      "step": 394
    },
    {
      "epoch": 0.15781062724730324,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0005,
      "loss": 1.1647,
      "step": 395
    },
    {
      "epoch": 0.15821014782261286,
      "grad_norm": 0.25,
      "learning_rate": 0.0005,
      "loss": 1.1523,
      "step": 396
    },
    {
      "epoch": 0.15860966839792248,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.1616,
      "step": 397
    },
    {
      "epoch": 0.15900918897323213,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0005,
      "loss": 1.1531,
      "step": 398
    },
    {
      "epoch": 0.15940870954854175,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0005,
      "loss": 1.1722,
      "step": 399
    },
    {
      "epoch": 0.15980823012385137,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0005,
      "loss": 1.1697,
      "step": 400
    },
    {
      "epoch": 0.16020775069916102,
      "grad_norm": 0.203125,
      "learning_rate": 0.0005,
      "loss": 1.1583,
      "step": 401
    },
    {
      "epoch": 0.16060727127447064,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0005,
      "loss": 1.1689,
      "step": 402
    },
    {
      "epoch": 0.16100679184978026,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0005,
      "loss": 1.1518,
      "step": 403
    },
    {
      "epoch": 0.1614063124250899,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.1612,
      "step": 404
    },
    {
      "epoch": 0.16180583300039952,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.1612,
      "step": 405
    },
    {
      "epoch": 0.16220535357570914,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.145,
      "step": 406
    },
    {
      "epoch": 0.1626048741510188,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.1456,
      "step": 407
    },
    {
      "epoch": 0.1630043947263284,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0005,
      "loss": 1.1394,
      "step": 408
    },
    {
      "epoch": 0.16340391530163803,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.1601,
      "step": 409
    },
    {
      "epoch": 0.16380343587694765,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005,
      "loss": 1.1604,
      "step": 410
    },
    {
      "epoch": 0.1642029564522573,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.1564,
      "step": 411
    },
    {
      "epoch": 0.16460247702756692,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0005,
      "loss": 1.1513,
      "step": 412
    },
    {
      "epoch": 0.16500199760287654,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.1528,
      "step": 413
    },
    {
      "epoch": 0.16540151817818619,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.141,
      "step": 414
    },
    {
      "epoch": 0.1658010387534958,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.1512,
      "step": 415
    },
    {
      "epoch": 0.16620055932880543,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.1418,
      "step": 416
    },
    {
      "epoch": 0.16660007990411507,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.1574,
      "step": 417
    },
    {
      "epoch": 0.1669996004794247,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.1474,
      "step": 418
    },
    {
      "epoch": 0.1673991210547343,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.1436,
      "step": 419
    },
    {
      "epoch": 0.16779864163004396,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.1517,
      "step": 420
    },
    {
      "epoch": 0.16819816220535358,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.1496,
      "step": 421
    },
    {
      "epoch": 0.1685976827806632,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.1506,
      "step": 422
    },
    {
      "epoch": 0.16899720335597282,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.135,
      "step": 423
    },
    {
      "epoch": 0.16939672393128247,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.1386,
      "step": 424
    },
    {
      "epoch": 0.1697962445065921,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0005,
      "loss": 1.1494,
      "step": 425
    },
    {
      "epoch": 0.1701957650819017,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.1435,
      "step": 426
    },
    {
      "epoch": 0.17059528565721135,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0005,
      "loss": 1.1372,
      "step": 427
    },
    {
      "epoch": 0.17099480623252097,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0005,
      "loss": 1.1506,
      "step": 428
    },
    {
      "epoch": 0.1713943268078306,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.1417,
      "step": 429
    },
    {
      "epoch": 0.17179384738314024,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0005,
      "loss": 1.1499,
      "step": 430
    },
    {
      "epoch": 0.17219336795844986,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.1516,
      "step": 431
    },
    {
      "epoch": 0.17259288853375948,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.1509,
      "step": 432
    },
    {
      "epoch": 0.17299240910906913,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.1441,
      "step": 433
    },
    {
      "epoch": 0.17339192968437875,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0005,
      "loss": 1.1348,
      "step": 434
    },
    {
      "epoch": 0.17379145025968837,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.1401,
      "step": 435
    },
    {
      "epoch": 0.17419097083499802,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.1448,
      "step": 436
    },
    {
      "epoch": 0.17459049141030764,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.1343,
      "step": 437
    },
    {
      "epoch": 0.17499001198561726,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.1418,
      "step": 438
    },
    {
      "epoch": 0.17538953256092688,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0005,
      "loss": 1.1444,
      "step": 439
    },
    {
      "epoch": 0.17578905313623652,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.1317,
      "step": 440
    },
    {
      "epoch": 0.17618857371154614,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.1473,
      "step": 441
    },
    {
      "epoch": 0.17658809428685576,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.1418,
      "step": 442
    },
    {
      "epoch": 0.1769876148621654,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0005,
      "loss": 1.1405,
      "step": 443
    },
    {
      "epoch": 0.17738713543747503,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.1554,
      "step": 444
    },
    {
      "epoch": 0.17778665601278465,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.1375,
      "step": 445
    },
    {
      "epoch": 0.1781861765880943,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0005,
      "loss": 1.1531,
      "step": 446
    },
    {
      "epoch": 0.17858569716340392,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.1341,
      "step": 447
    },
    {
      "epoch": 0.17898521773871354,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0005,
      "loss": 1.1401,
      "step": 448
    },
    {
      "epoch": 0.17938473831402318,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.1591,
      "step": 449
    },
    {
      "epoch": 0.1797842588893328,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0005,
      "loss": 1.1412,
      "step": 450
    },
    {
      "epoch": 0.18018377946464242,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.1372,
      "step": 451
    },
    {
      "epoch": 0.18058330003995204,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0005,
      "loss": 1.1486,
      "step": 452
    },
    {
      "epoch": 0.1809828206152617,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.1508,
      "step": 453
    },
    {
      "epoch": 0.1813823411905713,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.14,
      "step": 454
    },
    {
      "epoch": 0.18178186176588093,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.1421,
      "step": 455
    },
    {
      "epoch": 0.18218138234119058,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.1312,
      "step": 456
    },
    {
      "epoch": 0.1825809029165002,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0005,
      "loss": 1.1229,
      "step": 457
    },
    {
      "epoch": 0.18298042349180982,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.1395,
      "step": 458
    },
    {
      "epoch": 0.18337994406711947,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0005,
      "loss": 1.1456,
      "step": 459
    },
    {
      "epoch": 0.18377946464242909,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0005,
      "loss": 1.1404,
      "step": 460
    },
    {
      "epoch": 0.1841789852177387,
      "grad_norm": 0.5625,
      "learning_rate": 0.0005,
      "loss": 1.1426,
      "step": 461
    },
    {
      "epoch": 0.18457850579304835,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0005,
      "loss": 1.1609,
      "step": 462
    },
    {
      "epoch": 0.18497802636835797,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005,
      "loss": 1.1669,
      "step": 463
    },
    {
      "epoch": 0.1853775469436676,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0005,
      "loss": 1.1558,
      "step": 464
    },
    {
      "epoch": 0.18577706751897724,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005,
      "loss": 1.15,
      "step": 465
    },
    {
      "epoch": 0.18617658809428686,
      "grad_norm": 0.546875,
      "learning_rate": 0.0005,
      "loss": 1.1425,
      "step": 466
    },
    {
      "epoch": 0.18657610866959648,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0005,
      "loss": 1.1687,
      "step": 467
    },
    {
      "epoch": 0.1869756292449061,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0005,
      "loss": 1.1547,
      "step": 468
    },
    {
      "epoch": 0.18737514982021575,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005,
      "loss": 1.1379,
      "step": 469
    },
    {
      "epoch": 0.18777467039552537,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005,
      "loss": 1.1514,
      "step": 470
    },
    {
      "epoch": 0.188174190970835,
      "grad_norm": 0.3125,
      "learning_rate": 0.0005,
      "loss": 1.1585,
      "step": 471
    },
    {
      "epoch": 0.18857371154614463,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005,
      "loss": 1.153,
      "step": 472
    },
    {
      "epoch": 0.18897323212145425,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0005,
      "loss": 1.1497,
      "step": 473
    },
    {
      "epoch": 0.18937275269676387,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0005,
      "loss": 1.1651,
      "step": 474
    },
    {
      "epoch": 0.18977227327207352,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0005,
      "loss": 1.1518,
      "step": 475
    },
    {
      "epoch": 0.19017179384738314,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0005,
      "loss": 1.1351,
      "step": 476
    },
    {
      "epoch": 0.19057131442269276,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0005,
      "loss": 1.1641,
      "step": 477
    },
    {
      "epoch": 0.1909708349980024,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0005,
      "loss": 1.1333,
      "step": 478
    },
    {
      "epoch": 0.19137035557331203,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0005,
      "loss": 1.1447,
      "step": 479
    },
    {
      "epoch": 0.19176987614862165,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0005,
      "loss": 1.15,
      "step": 480
    },
    {
      "epoch": 0.1921693967239313,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0005,
      "loss": 1.125,
      "step": 481
    },
    {
      "epoch": 0.19256891729924092,
      "grad_norm": 0.203125,
      "learning_rate": 0.0005,
      "loss": 1.1514,
      "step": 482
    },
    {
      "epoch": 0.19296843787455054,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0005,
      "loss": 1.151,
      "step": 483
    },
    {
      "epoch": 0.19336795844986016,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0005,
      "loss": 1.1477,
      "step": 484
    },
    {
      "epoch": 0.1937674790251698,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.1395,
      "step": 485
    },
    {
      "epoch": 0.19416699960047942,
      "grad_norm": 0.171875,
      "learning_rate": 0.0005,
      "loss": 1.1447,
      "step": 486
    },
    {
      "epoch": 0.19456652017578904,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.1353,
      "step": 487
    },
    {
      "epoch": 0.1949660407510987,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.1354,
      "step": 488
    },
    {
      "epoch": 0.1953655613264083,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0005,
      "loss": 1.1455,
      "step": 489
    },
    {
      "epoch": 0.19576508190171793,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.1394,
      "step": 490
    },
    {
      "epoch": 0.19616460247702758,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.1361,
      "step": 491
    },
    {
      "epoch": 0.1965641230523372,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.1353,
      "step": 492
    },
    {
      "epoch": 0.19696364362764682,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.1408,
      "step": 493
    },
    {
      "epoch": 0.19736316420295646,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.1354,
      "step": 494
    },
    {
      "epoch": 0.19776268477826608,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.1524,
      "step": 495
    },
    {
      "epoch": 0.1981622053535757,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.1425,
      "step": 496
    },
    {
      "epoch": 0.19856172592888532,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.1277,
      "step": 497
    },
    {
      "epoch": 0.19896124650419497,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.1353,
      "step": 498
    },
    {
      "epoch": 0.1993607670795046,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.1432,
      "step": 499
    },
    {
      "epoch": 0.1997602876548142,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.1328,
      "step": 500
    },
    {
      "epoch": 0.20015980823012386,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.1392,
      "step": 501
    },
    {
      "epoch": 0.20055932880543348,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.1372,
      "step": 502
    },
    {
      "epoch": 0.2009588493807431,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.1478,
      "step": 503
    },
    {
      "epoch": 0.20135836995605275,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.1399,
      "step": 504
    },
    {
      "epoch": 0.20175789053136237,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.1361,
      "step": 505
    },
    {
      "epoch": 0.20215741110667199,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.1413,
      "step": 506
    },
    {
      "epoch": 0.20255693168198163,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.1247,
      "step": 507
    },
    {
      "epoch": 0.20295645225729125,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0005,
      "loss": 1.1337,
      "step": 508
    },
    {
      "epoch": 0.20335597283260087,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.1325,
      "step": 509
    },
    {
      "epoch": 0.20375549340791052,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.1347,
      "step": 510
    },
    {
      "epoch": 0.20415501398322014,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.1469,
      "step": 511
    },
    {
      "epoch": 0.20455453455852976,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.1347,
      "step": 512
    },
    {
      "epoch": 0.20495405513383938,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.1358,
      "step": 513
    },
    {
      "epoch": 0.20535357570914903,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0005,
      "loss": 1.1423,
      "step": 514
    },
    {
      "epoch": 0.20575309628445865,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0005,
      "loss": 1.1375,
      "step": 515
    },
    {
      "epoch": 0.20615261685976827,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.131,
      "step": 516
    },
    {
      "epoch": 0.20655213743507791,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005,
      "loss": 1.1334,
      "step": 517
    },
    {
      "epoch": 0.20695165801038753,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0005,
      "loss": 1.1287,
      "step": 518
    },
    {
      "epoch": 0.20735117858569715,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.1275,
      "step": 519
    },
    {
      "epoch": 0.2077506991610068,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0005,
      "loss": 1.1333,
      "step": 520
    },
    {
      "epoch": 0.20815021973631642,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.1339,
      "step": 521
    },
    {
      "epoch": 0.20854974031162604,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0005,
      "loss": 1.1218,
      "step": 522
    },
    {
      "epoch": 0.2089492608869357,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.1369,
      "step": 523
    },
    {
      "epoch": 0.2093487814622453,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.1356,
      "step": 524
    },
    {
      "epoch": 0.20974830203755493,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0005,
      "loss": 1.1366,
      "step": 525
    },
    {
      "epoch": 0.21014782261286455,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.1324,
      "step": 526
    },
    {
      "epoch": 0.2105473431881742,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.1347,
      "step": 527
    },
    {
      "epoch": 0.21094686376348382,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.1422,
      "step": 528
    },
    {
      "epoch": 0.21134638433879344,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.1349,
      "step": 529
    },
    {
      "epoch": 0.21174590491410308,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.1324,
      "step": 530
    },
    {
      "epoch": 0.2121454254894127,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.1373,
      "step": 531
    },
    {
      "epoch": 0.21254494606472232,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.1336,
      "step": 532
    },
    {
      "epoch": 0.21294446664003197,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.1275,
      "step": 533
    },
    {
      "epoch": 0.2133439872153416,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0005,
      "loss": 1.1243,
      "step": 534
    },
    {
      "epoch": 0.2137435077906512,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0005,
      "loss": 1.1339,
      "step": 535
    },
    {
      "epoch": 0.21414302836596086,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0005,
      "loss": 1.1393,
      "step": 536
    },
    {
      "epoch": 0.21454254894127048,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0005,
      "loss": 1.1349,
      "step": 537
    },
    {
      "epoch": 0.2149420695165801,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0005,
      "loss": 1.144,
      "step": 538
    },
    {
      "epoch": 0.21534159009188975,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.1174,
      "step": 539
    },
    {
      "epoch": 0.21574111066719937,
      "grad_norm": 0.46875,
      "learning_rate": 0.0005,
      "loss": 1.1477,
      "step": 540
    },
    {
      "epoch": 0.21614063124250898,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0005,
      "loss": 1.1354,
      "step": 541
    },
    {
      "epoch": 0.2165401518178186,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0005,
      "loss": 1.1227,
      "step": 542
    },
    {
      "epoch": 0.21693967239312825,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0005,
      "loss": 1.1354,
      "step": 543
    },
    {
      "epoch": 0.21733919296843787,
      "grad_norm": 0.578125,
      "learning_rate": 0.0005,
      "loss": 1.1385,
      "step": 544
    },
    {
      "epoch": 0.2177387135437475,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0005,
      "loss": 1.1243,
      "step": 545
    },
    {
      "epoch": 0.21813823411905714,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0005,
      "loss": 1.1406,
      "step": 546
    },
    {
      "epoch": 0.21853775469436676,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0005,
      "loss": 1.1442,
      "step": 547
    },
    {
      "epoch": 0.21893727526967638,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005,
      "loss": 1.1316,
      "step": 548
    },
    {
      "epoch": 0.21933679584498603,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.1311,
      "step": 549
    },
    {
      "epoch": 0.21973631642029565,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0005,
      "loss": 1.1231,
      "step": 550
    },
    {
      "epoch": 0.22013583699560527,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.1372,
      "step": 551
    },
    {
      "epoch": 0.2205353575709149,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005,
      "loss": 1.1384,
      "step": 552
    },
    {
      "epoch": 0.22093487814622453,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0005,
      "loss": 1.1391,
      "step": 553
    },
    {
      "epoch": 0.22133439872153415,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0005,
      "loss": 1.1391,
      "step": 554
    },
    {
      "epoch": 0.2217339192968438,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0005,
      "loss": 1.1501,
      "step": 555
    },
    {
      "epoch": 0.22213343987215342,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0005,
      "loss": 1.1299,
      "step": 556
    },
    {
      "epoch": 0.22253296044746304,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0005,
      "loss": 1.1356,
      "step": 557
    },
    {
      "epoch": 0.22293248102277266,
      "grad_norm": 0.46875,
      "learning_rate": 0.0005,
      "loss": 1.1489,
      "step": 558
    },
    {
      "epoch": 0.2233320015980823,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0005,
      "loss": 1.1265,
      "step": 559
    },
    {
      "epoch": 0.22373152217339193,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005,
      "loss": 1.135,
      "step": 560
    },
    {
      "epoch": 0.22413104274870155,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0005,
      "loss": 1.1339,
      "step": 561
    },
    {
      "epoch": 0.2245305633240112,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0005,
      "loss": 1.1329,
      "step": 562
    },
    {
      "epoch": 0.22493008389932082,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0005,
      "loss": 1.1349,
      "step": 563
    },
    {
      "epoch": 0.22532960447463043,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0005,
      "loss": 1.1348,
      "step": 564
    },
    {
      "epoch": 0.22572912504994008,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005,
      "loss": 1.1405,
      "step": 565
    },
    {
      "epoch": 0.2261286456252497,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.1311,
      "step": 566
    },
    {
      "epoch": 0.22652816620055932,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0005,
      "loss": 1.1309,
      "step": 567
    },
    {
      "epoch": 0.22692768677586897,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0005,
      "loss": 1.1266,
      "step": 568
    },
    {
      "epoch": 0.2273272073511786,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0005,
      "loss": 1.1294,
      "step": 569
    },
    {
      "epoch": 0.2277267279264882,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0005,
      "loss": 1.1306,
      "step": 570
    },
    {
      "epoch": 0.22812624850179783,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.1377,
      "step": 571
    },
    {
      "epoch": 0.22852576907710748,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.1154,
      "step": 572
    },
    {
      "epoch": 0.2289252896524171,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.1247,
      "step": 573
    },
    {
      "epoch": 0.22932481022772672,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0005,
      "loss": 1.135,
      "step": 574
    },
    {
      "epoch": 0.22972433080303636,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.1334,
      "step": 575
    },
    {
      "epoch": 0.23012385137834598,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0005,
      "loss": 1.1201,
      "step": 576
    },
    {
      "epoch": 0.2305233719536556,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.13,
      "step": 577
    },
    {
      "epoch": 0.23092289252896525,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0005,
      "loss": 1.1216,
      "step": 578
    },
    {
      "epoch": 0.23132241310427487,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.132,
      "step": 579
    },
    {
      "epoch": 0.2317219336795845,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.1421,
      "step": 580
    },
    {
      "epoch": 0.23212145425489414,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.1248,
      "step": 581
    },
    {
      "epoch": 0.23252097483020376,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.1282,
      "step": 582
    },
    {
      "epoch": 0.23292049540551338,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.1232,
      "step": 583
    },
    {
      "epoch": 0.23332001598082303,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.13,
      "step": 584
    },
    {
      "epoch": 0.23371953655613265,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.1497,
      "step": 585
    },
    {
      "epoch": 0.23411905713144227,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.1269,
      "step": 586
    },
    {
      "epoch": 0.23451857770675189,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.1265,
      "step": 587
    },
    {
      "epoch": 0.23491809828206153,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0005,
      "loss": 1.1292,
      "step": 588
    },
    {
      "epoch": 0.23531761885737115,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.1211,
      "step": 589
    },
    {
      "epoch": 0.23571713943268077,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0005,
      "loss": 1.1391,
      "step": 590
    },
    {
      "epoch": 0.23611666000799042,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.1245,
      "step": 591
    },
    {
      "epoch": 0.23651618058330004,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.1238,
      "step": 592
    },
    {
      "epoch": 0.23691570115860966,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005,
      "loss": 1.1251,
      "step": 593
    },
    {
      "epoch": 0.2373152217339193,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.123,
      "step": 594
    },
    {
      "epoch": 0.23771474230922893,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.141,
      "step": 595
    },
    {
      "epoch": 0.23811426288453855,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0005,
      "loss": 1.1366,
      "step": 596
    },
    {
      "epoch": 0.2385137834598482,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.1364,
      "step": 597
    },
    {
      "epoch": 0.23891330403515781,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.1281,
      "step": 598
    },
    {
      "epoch": 0.23931282461046743,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.1326,
      "step": 599
    },
    {
      "epoch": 0.23971234518577705,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.1316,
      "step": 600
    },
    {
      "epoch": 0.2401118657610867,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.1346,
      "step": 601
    },
    {
      "epoch": 0.24051138633639632,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.1267,
      "step": 602
    },
    {
      "epoch": 0.24091090691170594,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.1284,
      "step": 603
    },
    {
      "epoch": 0.2413104274870156,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.1287,
      "step": 604
    },
    {
      "epoch": 0.2417099480623252,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.131,
      "step": 605
    },
    {
      "epoch": 0.24210946863763483,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.1251,
      "step": 606
    },
    {
      "epoch": 0.24250898921294448,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.1271,
      "step": 607
    },
    {
      "epoch": 0.2429085097882541,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.1347,
      "step": 608
    },
    {
      "epoch": 0.24330803036356372,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.1315,
      "step": 609
    },
    {
      "epoch": 0.24370755093887336,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.1335,
      "step": 610
    },
    {
      "epoch": 0.24410707151418298,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.1253,
      "step": 611
    },
    {
      "epoch": 0.2445065920894926,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.1233,
      "step": 612
    },
    {
      "epoch": 0.24490611266480225,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.1331,
      "step": 613
    },
    {
      "epoch": 0.24530563324011187,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.1334,
      "step": 614
    },
    {
      "epoch": 0.2457051538154215,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.1246,
      "step": 615
    },
    {
      "epoch": 0.2461046743907311,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.1244,
      "step": 616
    },
    {
      "epoch": 0.24650419496604076,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.1424,
      "step": 617
    },
    {
      "epoch": 0.24690371554135038,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.1149,
      "step": 618
    },
    {
      "epoch": 0.24730323611666,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0005,
      "loss": 1.1339,
      "step": 619
    },
    {
      "epoch": 0.24770275669196964,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0005,
      "loss": 1.1354,
      "step": 620
    },
    {
      "epoch": 0.24810227726727926,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0005,
      "loss": 1.1306,
      "step": 621
    },
    {
      "epoch": 0.24850179784258888,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.1264,
      "step": 622
    },
    {
      "epoch": 0.24890131841789853,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.1187,
      "step": 623
    },
    {
      "epoch": 0.24930083899320815,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.1114,
      "step": 624
    },
    {
      "epoch": 0.24970035956851777,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0005,
      "loss": 1.1332,
      "step": 625
    },
    {
      "epoch": 0.2500998801438274,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.1336,
      "step": 626
    },
    {
      "epoch": 0.250499400719137,
      "grad_norm": 0.25,
      "learning_rate": 0.0005,
      "loss": 1.1356,
      "step": 627
    },
    {
      "epoch": 0.2508989212944467,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0005,
      "loss": 1.1243,
      "step": 628
    },
    {
      "epoch": 0.2512984418697563,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0005,
      "loss": 1.1232,
      "step": 629
    },
    {
      "epoch": 0.2516979624450659,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.1225,
      "step": 630
    },
    {
      "epoch": 0.25209748302037555,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0005,
      "loss": 1.1292,
      "step": 631
    },
    {
      "epoch": 0.25249700359568517,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.121,
      "step": 632
    },
    {
      "epoch": 0.2528965241709948,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.1255,
      "step": 633
    },
    {
      "epoch": 0.25329604474630446,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.1247,
      "step": 634
    },
    {
      "epoch": 0.2536955653216141,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.1361,
      "step": 635
    },
    {
      "epoch": 0.2540950858969237,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0005,
      "loss": 1.1327,
      "step": 636
    },
    {
      "epoch": 0.2544946064722333,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.134,
      "step": 637
    },
    {
      "epoch": 0.25489412704754294,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.1243,
      "step": 638
    },
    {
      "epoch": 0.25529364762285256,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0005,
      "loss": 1.1283,
      "step": 639
    },
    {
      "epoch": 0.2556931681981622,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.1386,
      "step": 640
    },
    {
      "epoch": 0.25609268877347185,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0005,
      "loss": 1.1331,
      "step": 641
    },
    {
      "epoch": 0.2564922093487815,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0005,
      "loss": 1.1305,
      "step": 642
    },
    {
      "epoch": 0.2568917299240911,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.1259,
      "step": 643
    },
    {
      "epoch": 0.2572912504994007,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0005,
      "loss": 1.1218,
      "step": 644
    },
    {
      "epoch": 0.25769077107471033,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.1406,
      "step": 645
    },
    {
      "epoch": 0.25809029165001995,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.1359,
      "step": 646
    },
    {
      "epoch": 0.25848981222532963,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.1266,
      "step": 647
    },
    {
      "epoch": 0.25888933280063925,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.1351,
      "step": 648
    },
    {
      "epoch": 0.25928885337594887,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0005,
      "loss": 1.132,
      "step": 649
    },
    {
      "epoch": 0.2596883739512585,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.1172,
      "step": 650
    },
    {
      "epoch": 0.2600878945265681,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.1223,
      "step": 651
    },
    {
      "epoch": 0.26048741510187773,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0005,
      "loss": 1.1121,
      "step": 652
    },
    {
      "epoch": 0.26088693567718735,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.1363,
      "step": 653
    },
    {
      "epoch": 0.261286456252497,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.127,
      "step": 654
    },
    {
      "epoch": 0.26168597682780664,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.1282,
      "step": 655
    },
    {
      "epoch": 0.26208549740311626,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.1261,
      "step": 656
    },
    {
      "epoch": 0.2624850179784259,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.1165,
      "step": 657
    },
    {
      "epoch": 0.2628845385537355,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.1372,
      "step": 658
    },
    {
      "epoch": 0.2632840591290451,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.124,
      "step": 659
    },
    {
      "epoch": 0.2636835797043548,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.1371,
      "step": 660
    },
    {
      "epoch": 0.2640831002796644,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.1271,
      "step": 661
    },
    {
      "epoch": 0.26448262085497404,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.1355,
      "step": 662
    },
    {
      "epoch": 0.26488214143028366,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.124,
      "step": 663
    },
    {
      "epoch": 0.2652816620055933,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.115,
      "step": 664
    },
    {
      "epoch": 0.2656811825809029,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.1229,
      "step": 665
    },
    {
      "epoch": 0.2660807031562126,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.1194,
      "step": 666
    },
    {
      "epoch": 0.2664802237315222,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.1218,
      "step": 667
    },
    {
      "epoch": 0.2668797443068318,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.1148,
      "step": 668
    },
    {
      "epoch": 0.26727926488214143,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.1313,
      "step": 669
    },
    {
      "epoch": 0.26767878545745105,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.1186,
      "step": 670
    },
    {
      "epoch": 0.26807830603276067,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.1332,
      "step": 671
    },
    {
      "epoch": 0.2684778266080703,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.1139,
      "step": 672
    },
    {
      "epoch": 0.26887734718337997,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.1394,
      "step": 673
    },
    {
      "epoch": 0.2692768677586896,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.1202,
      "step": 674
    },
    {
      "epoch": 0.2696763883339992,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.1184,
      "step": 675
    },
    {
      "epoch": 0.2700759089093088,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.1266,
      "step": 676
    },
    {
      "epoch": 0.27047542948461845,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.1284,
      "step": 677
    },
    {
      "epoch": 0.27087495005992807,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.1219,
      "step": 678
    },
    {
      "epoch": 0.27127447063523774,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.1241,
      "step": 679
    },
    {
      "epoch": 0.27167399121054736,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0005,
      "loss": 1.1102,
      "step": 680
    },
    {
      "epoch": 0.272073511785857,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0005,
      "loss": 1.1313,
      "step": 681
    },
    {
      "epoch": 0.2724730323611666,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0005,
      "loss": 1.1362,
      "step": 682
    },
    {
      "epoch": 0.2728725529364762,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.1326,
      "step": 683
    },
    {
      "epoch": 0.27327207351178584,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0005,
      "loss": 1.1277,
      "step": 684
    },
    {
      "epoch": 0.27367159408709546,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005,
      "loss": 1.1249,
      "step": 685
    },
    {
      "epoch": 0.27407111466240514,
      "grad_norm": 0.171875,
      "learning_rate": 0.0005,
      "loss": 1.1239,
      "step": 686
    },
    {
      "epoch": 0.27447063523771476,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.1283,
      "step": 687
    },
    {
      "epoch": 0.2748701558130244,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0005,
      "loss": 1.1168,
      "step": 688
    },
    {
      "epoch": 0.275269676388334,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.1284,
      "step": 689
    },
    {
      "epoch": 0.2756691969636436,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.1296,
      "step": 690
    },
    {
      "epoch": 0.27606871753895323,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.1217,
      "step": 691
    },
    {
      "epoch": 0.2764682381142629,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.1094,
      "step": 692
    },
    {
      "epoch": 0.27686775868957253,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.1169,
      "step": 693
    },
    {
      "epoch": 0.27726727926488215,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.1339,
      "step": 694
    },
    {
      "epoch": 0.27766679984019177,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.1377,
      "step": 695
    },
    {
      "epoch": 0.2780663204155014,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.1182,
      "step": 696
    },
    {
      "epoch": 0.278465840990811,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.1103,
      "step": 697
    },
    {
      "epoch": 0.27886536156612063,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0005,
      "loss": 1.1235,
      "step": 698
    },
    {
      "epoch": 0.2792648821414303,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0005,
      "loss": 1.128,
      "step": 699
    },
    {
      "epoch": 0.2796644027167399,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.126,
      "step": 700
    },
    {
      "epoch": 0.28006392329204954,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.1206,
      "step": 701
    },
    {
      "epoch": 0.28046344386735916,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.121,
      "step": 702
    },
    {
      "epoch": 0.2808629644426688,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.1101,
      "step": 703
    },
    {
      "epoch": 0.2812624850179784,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.1231,
      "step": 704
    },
    {
      "epoch": 0.2816620055932881,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.1226,
      "step": 705
    },
    {
      "epoch": 0.2820615261685977,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.1168,
      "step": 706
    },
    {
      "epoch": 0.2824610467439073,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.1158,
      "step": 707
    },
    {
      "epoch": 0.28286056731921694,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.1119,
      "step": 708
    },
    {
      "epoch": 0.28326008789452656,
      "grad_norm": 0.21875,
      "learning_rate": 0.0005,
      "loss": 1.1213,
      "step": 709
    },
    {
      "epoch": 0.2836596084698362,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.1258,
      "step": 710
    },
    {
      "epoch": 0.2840591290451458,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.1153,
      "step": 711
    },
    {
      "epoch": 0.2844586496204555,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.1101,
      "step": 712
    },
    {
      "epoch": 0.2848581701957651,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.1188,
      "step": 713
    },
    {
      "epoch": 0.2852576907710747,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.1233,
      "step": 714
    },
    {
      "epoch": 0.28565721134638433,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.1138,
      "step": 715
    },
    {
      "epoch": 0.28605673192169395,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.1177,
      "step": 716
    },
    {
      "epoch": 0.28645625249700357,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0005,
      "loss": 1.1249,
      "step": 717
    },
    {
      "epoch": 0.28685577307231325,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0005,
      "loss": 1.1251,
      "step": 718
    },
    {
      "epoch": 0.28725529364762287,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005,
      "loss": 1.1244,
      "step": 719
    },
    {
      "epoch": 0.2876548142229325,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005,
      "loss": 1.1146,
      "step": 720
    },
    {
      "epoch": 0.2880543347982421,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0005,
      "loss": 1.1307,
      "step": 721
    },
    {
      "epoch": 0.2884538553735517,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.1092,
      "step": 722
    },
    {
      "epoch": 0.28885337594886135,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0005,
      "loss": 1.1134,
      "step": 723
    },
    {
      "epoch": 0.289252896524171,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.1236,
      "step": 724
    },
    {
      "epoch": 0.28965241709948064,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.1193,
      "step": 725
    },
    {
      "epoch": 0.29005193767479026,
      "grad_norm": 0.203125,
      "learning_rate": 0.0005,
      "loss": 1.1269,
      "step": 726
    },
    {
      "epoch": 0.2904514582500999,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.1269,
      "step": 727
    },
    {
      "epoch": 0.2908509788254095,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.1203,
      "step": 728
    },
    {
      "epoch": 0.2912504994007191,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.1312,
      "step": 729
    },
    {
      "epoch": 0.29165001997602874,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.1167,
      "step": 730
    },
    {
      "epoch": 0.2920495405513384,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0005,
      "loss": 1.1268,
      "step": 731
    },
    {
      "epoch": 0.29244906112664804,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005,
      "loss": 1.1221,
      "step": 732
    },
    {
      "epoch": 0.29284858170195766,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0005,
      "loss": 1.1234,
      "step": 733
    },
    {
      "epoch": 0.2932481022772673,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0005,
      "loss": 1.1292,
      "step": 734
    },
    {
      "epoch": 0.2936476228525769,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005,
      "loss": 1.1626,
      "step": 735
    },
    {
      "epoch": 0.2940471434278865,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005,
      "loss": 1.1371,
      "step": 736
    },
    {
      "epoch": 0.2944466640031962,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0005,
      "loss": 1.138,
      "step": 737
    },
    {
      "epoch": 0.2948461845785058,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0005,
      "loss": 1.1269,
      "step": 738
    },
    {
      "epoch": 0.29524570515381543,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0005,
      "loss": 1.1353,
      "step": 739
    },
    {
      "epoch": 0.29564522572912505,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0005,
      "loss": 1.1402,
      "step": 740
    },
    {
      "epoch": 0.29604474630443467,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0005,
      "loss": 1.1354,
      "step": 741
    },
    {
      "epoch": 0.2964442668797443,
      "grad_norm": 0.203125,
      "learning_rate": 0.0005,
      "loss": 1.1138,
      "step": 742
    },
    {
      "epoch": 0.2968437874550539,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0005,
      "loss": 1.1408,
      "step": 743
    },
    {
      "epoch": 0.2972433080303636,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0005,
      "loss": 1.1339,
      "step": 744
    },
    {
      "epoch": 0.2976428286056732,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0005,
      "loss": 1.12,
      "step": 745
    },
    {
      "epoch": 0.2980423491809828,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.1109,
      "step": 746
    },
    {
      "epoch": 0.29844186975629244,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.1288,
      "step": 747
    },
    {
      "epoch": 0.29884139033160206,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.1201,
      "step": 748
    },
    {
      "epoch": 0.2992409109069117,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.1236,
      "step": 749
    },
    {
      "epoch": 0.29964043148222136,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.1252,
      "step": 750
    },
    {
      "epoch": 0.300039952057531,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.1265,
      "step": 751
    },
    {
      "epoch": 0.3004394726328406,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.1172,
      "step": 752
    },
    {
      "epoch": 0.3008389932081502,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.1404,
      "step": 753
    },
    {
      "epoch": 0.30123851378345984,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.1316,
      "step": 754
    },
    {
      "epoch": 0.30163803435876946,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.1273,
      "step": 755
    },
    {
      "epoch": 0.3020375549340791,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0005,
      "loss": 1.1202,
      "step": 756
    },
    {
      "epoch": 0.30243707550938875,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.1228,
      "step": 757
    },
    {
      "epoch": 0.3028365960846984,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.119,
      "step": 758
    },
    {
      "epoch": 0.303236116660008,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.119,
      "step": 759
    },
    {
      "epoch": 0.3036356372353176,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.1221,
      "step": 760
    },
    {
      "epoch": 0.30403515781062723,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.1136,
      "step": 761
    },
    {
      "epoch": 0.30443467838593685,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.1258,
      "step": 762
    },
    {
      "epoch": 0.3048341989612465,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.1278,
      "step": 763
    },
    {
      "epoch": 0.30523371953655615,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.1177,
      "step": 764
    },
    {
      "epoch": 0.30563324011186577,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.1233,
      "step": 765
    },
    {
      "epoch": 0.3060327606871754,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.1077,
      "step": 766
    },
    {
      "epoch": 0.306432281262485,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.1229,
      "step": 767
    },
    {
      "epoch": 0.3068318018377946,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.1112,
      "step": 768
    },
    {
      "epoch": 0.3072313224131043,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.1127,
      "step": 769
    },
    {
      "epoch": 0.3076308429884139,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.1207,
      "step": 770
    },
    {
      "epoch": 0.30803036356372354,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.1061,
      "step": 771
    },
    {
      "epoch": 0.30842988413903316,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.1155,
      "step": 772
    },
    {
      "epoch": 0.3088294047143428,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.1171,
      "step": 773
    },
    {
      "epoch": 0.3092289252896524,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.1338,
      "step": 774
    },
    {
      "epoch": 0.309628445864962,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.1294,
      "step": 775
    },
    {
      "epoch": 0.3100279664402717,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.1079,
      "step": 776
    },
    {
      "epoch": 0.3104274870155813,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.1213,
      "step": 777
    },
    {
      "epoch": 0.31082700759089094,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.1129,
      "step": 778
    },
    {
      "epoch": 0.31122652816620056,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.1111,
      "step": 779
    },
    {
      "epoch": 0.3116260487415102,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0005,
      "loss": 1.1093,
      "step": 780
    },
    {
      "epoch": 0.3120255693168198,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.1203,
      "step": 781
    },
    {
      "epoch": 0.31242508989212947,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0005,
      "loss": 1.112,
      "step": 782
    },
    {
      "epoch": 0.3128246104674391,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.1165,
      "step": 783
    },
    {
      "epoch": 0.3132241310427487,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.1284,
      "step": 784
    },
    {
      "epoch": 0.31362365161805833,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.1089,
      "step": 785
    },
    {
      "epoch": 0.31402317219336795,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.1198,
      "step": 786
    },
    {
      "epoch": 0.31442269276867757,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.128,
      "step": 787
    },
    {
      "epoch": 0.3148222133439872,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.1186,
      "step": 788
    },
    {
      "epoch": 0.31522173391929686,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.1155,
      "step": 789
    },
    {
      "epoch": 0.3156212544946065,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.1202,
      "step": 790
    },
    {
      "epoch": 0.3160207750699161,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.1068,
      "step": 791
    },
    {
      "epoch": 0.3164202956452257,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.1252,
      "step": 792
    },
    {
      "epoch": 0.31681981622053534,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.1064,
      "step": 793
    },
    {
      "epoch": 0.31721933679584496,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.1271,
      "step": 794
    },
    {
      "epoch": 0.31761885737115464,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.113,
      "step": 795
    },
    {
      "epoch": 0.31801837794646426,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.1096,
      "step": 796
    },
    {
      "epoch": 0.3184178985217739,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.1106,
      "step": 797
    },
    {
      "epoch": 0.3188174190970835,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.1102,
      "step": 798
    },
    {
      "epoch": 0.3192169396723931,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.1239,
      "step": 799
    },
    {
      "epoch": 0.31961646024770274,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.1295,
      "step": 800
    },
    {
      "epoch": 0.32001598082301236,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.1133,
      "step": 801
    },
    {
      "epoch": 0.32041550139832203,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.1186,
      "step": 802
    },
    {
      "epoch": 0.32081502197363165,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.1266,
      "step": 803
    },
    {
      "epoch": 0.3212145425489413,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.1072,
      "step": 804
    },
    {
      "epoch": 0.3216140631242509,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.113,
      "step": 805
    },
    {
      "epoch": 0.3220135836995605,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.1033,
      "step": 806
    },
    {
      "epoch": 0.32241310427487013,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.1226,
      "step": 807
    },
    {
      "epoch": 0.3228126248501798,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.1169,
      "step": 808
    },
    {
      "epoch": 0.3232121454254894,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.1078,
      "step": 809
    },
    {
      "epoch": 0.32361166600079905,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.1135,
      "step": 810
    },
    {
      "epoch": 0.32401118657610867,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.1145,
      "step": 811
    },
    {
      "epoch": 0.3244107071514183,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.1243,
      "step": 812
    },
    {
      "epoch": 0.3248102277267279,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.121,
      "step": 813
    },
    {
      "epoch": 0.3252097483020376,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.1188,
      "step": 814
    },
    {
      "epoch": 0.3256092688773472,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.1085,
      "step": 815
    },
    {
      "epoch": 0.3260087894526568,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0005,
      "loss": 1.1207,
      "step": 816
    },
    {
      "epoch": 0.32640831002796644,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.1218,
      "step": 817
    },
    {
      "epoch": 0.32680783060327606,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.1157,
      "step": 818
    },
    {
      "epoch": 0.3272073511785857,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.1155,
      "step": 819
    },
    {
      "epoch": 0.3276068717538953,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.108,
      "step": 820
    },
    {
      "epoch": 0.328006392329205,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.1131,
      "step": 821
    },
    {
      "epoch": 0.3284059129045146,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.1157,
      "step": 822
    },
    {
      "epoch": 0.3288054334798242,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.1193,
      "step": 823
    },
    {
      "epoch": 0.32920495405513384,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.114,
      "step": 824
    },
    {
      "epoch": 0.32960447463044346,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.1106,
      "step": 825
    },
    {
      "epoch": 0.3300039952057531,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.1167,
      "step": 826
    },
    {
      "epoch": 0.33040351578106275,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.1092,
      "step": 827
    },
    {
      "epoch": 0.33080303635637237,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.1154,
      "step": 828
    },
    {
      "epoch": 0.331202556931682,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.1141,
      "step": 829
    },
    {
      "epoch": 0.3316020775069916,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.1112,
      "step": 830
    },
    {
      "epoch": 0.33200159808230123,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.1048,
      "step": 831
    },
    {
      "epoch": 0.33240111865761085,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.1215,
      "step": 832
    },
    {
      "epoch": 0.33280063923292047,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.122,
      "step": 833
    },
    {
      "epoch": 0.33320015980823015,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.1039,
      "step": 834
    },
    {
      "epoch": 0.33359968038353977,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.1194,
      "step": 835
    },
    {
      "epoch": 0.3339992009588494,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.1186,
      "step": 836
    },
    {
      "epoch": 0.334398721534159,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.1022,
      "step": 837
    },
    {
      "epoch": 0.3347982421094686,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.1134,
      "step": 838
    },
    {
      "epoch": 0.33519776268477824,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.12,
      "step": 839
    },
    {
      "epoch": 0.3355972832600879,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0934,
      "step": 840
    },
    {
      "epoch": 0.33599680383539754,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.115,
      "step": 841
    },
    {
      "epoch": 0.33639632441070716,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.1097,
      "step": 842
    },
    {
      "epoch": 0.3367958449860168,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.1192,
      "step": 843
    },
    {
      "epoch": 0.3371953655613264,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.1116,
      "step": 844
    },
    {
      "epoch": 0.337594886136636,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0995,
      "step": 845
    },
    {
      "epoch": 0.33799440671194564,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.1177,
      "step": 846
    },
    {
      "epoch": 0.3383939272872553,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.1161,
      "step": 847
    },
    {
      "epoch": 0.33879344786256493,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.1124,
      "step": 848
    },
    {
      "epoch": 0.33919296843787455,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.1089,
      "step": 849
    },
    {
      "epoch": 0.3395924890131842,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.1224,
      "step": 850
    },
    {
      "epoch": 0.3399920095884938,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.1175,
      "step": 851
    },
    {
      "epoch": 0.3403915301638034,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.1012,
      "step": 852
    },
    {
      "epoch": 0.3407910507391131,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.1187,
      "step": 853
    },
    {
      "epoch": 0.3411905713144227,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.1145,
      "step": 854
    },
    {
      "epoch": 0.34159009188973233,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.1097,
      "step": 855
    },
    {
      "epoch": 0.34198961246504195,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.1213,
      "step": 856
    },
    {
      "epoch": 0.34238913304035157,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0005,
      "loss": 1.1092,
      "step": 857
    },
    {
      "epoch": 0.3427886536156612,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0005,
      "loss": 1.12,
      "step": 858
    },
    {
      "epoch": 0.34318817419097086,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.1124,
      "step": 859
    },
    {
      "epoch": 0.3435876947662805,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005,
      "loss": 1.1176,
      "step": 860
    },
    {
      "epoch": 0.3439872153415901,
      "grad_norm": 0.4375,
      "learning_rate": 0.0005,
      "loss": 1.108,
      "step": 861
    },
    {
      "epoch": 0.3443867359168997,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0005,
      "loss": 1.1174,
      "step": 862
    },
    {
      "epoch": 0.34478625649220934,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0005,
      "loss": 1.1245,
      "step": 863
    },
    {
      "epoch": 0.34518577706751896,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005,
      "loss": 1.1434,
      "step": 864
    },
    {
      "epoch": 0.3455852976428286,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0005,
      "loss": 1.1176,
      "step": 865
    },
    {
      "epoch": 0.34598481821813826,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0005,
      "loss": 1.1277,
      "step": 866
    },
    {
      "epoch": 0.3463843387934479,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0005,
      "loss": 1.1293,
      "step": 867
    },
    {
      "epoch": 0.3467838593687575,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0005,
      "loss": 1.1049,
      "step": 868
    },
    {
      "epoch": 0.3471833799440671,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0005,
      "loss": 1.136,
      "step": 869
    },
    {
      "epoch": 0.34758290051937674,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0005,
      "loss": 1.1131,
      "step": 870
    },
    {
      "epoch": 0.34798242109468636,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0005,
      "loss": 1.1289,
      "step": 871
    },
    {
      "epoch": 0.34838194166999603,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0005,
      "loss": 1.1295,
      "step": 872
    },
    {
      "epoch": 0.34878146224530565,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0005,
      "loss": 1.117,
      "step": 873
    },
    {
      "epoch": 0.34918098282061527,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.114,
      "step": 874
    },
    {
      "epoch": 0.3495805033959249,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.1288,
      "step": 875
    },
    {
      "epoch": 0.3499800239712345,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.115,
      "step": 876
    },
    {
      "epoch": 0.35037954454654413,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0005,
      "loss": 1.1176,
      "step": 877
    },
    {
      "epoch": 0.35077906512185375,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.1187,
      "step": 878
    },
    {
      "epoch": 0.3511785856971634,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0005,
      "loss": 1.1207,
      "step": 879
    },
    {
      "epoch": 0.35157810627247305,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.1152,
      "step": 880
    },
    {
      "epoch": 0.35197762684778267,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.116,
      "step": 881
    },
    {
      "epoch": 0.3523771474230923,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.1172,
      "step": 882
    },
    {
      "epoch": 0.3527766679984019,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.1058,
      "step": 883
    },
    {
      "epoch": 0.3531761885737115,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.1245,
      "step": 884
    },
    {
      "epoch": 0.3535757091490212,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.1213,
      "step": 885
    },
    {
      "epoch": 0.3539752297243308,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.1153,
      "step": 886
    },
    {
      "epoch": 0.35437475029964044,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.1188,
      "step": 887
    },
    {
      "epoch": 0.35477427087495006,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.1298,
      "step": 888
    },
    {
      "epoch": 0.3551737914502597,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.1091,
      "step": 889
    },
    {
      "epoch": 0.3555733120255693,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.1128,
      "step": 890
    },
    {
      "epoch": 0.3559728326008789,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.104,
      "step": 891
    },
    {
      "epoch": 0.3563723531761886,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.1159,
      "step": 892
    },
    {
      "epoch": 0.3567718737514982,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.1067,
      "step": 893
    },
    {
      "epoch": 0.35717139432680783,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0005,
      "loss": 1.1072,
      "step": 894
    },
    {
      "epoch": 0.35757091490211745,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.1018,
      "step": 895
    },
    {
      "epoch": 0.3579704354774271,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0005,
      "loss": 1.1174,
      "step": 896
    },
    {
      "epoch": 0.3583699560527367,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.1152,
      "step": 897
    },
    {
      "epoch": 0.35876947662804637,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0005,
      "loss": 1.11,
      "step": 898
    },
    {
      "epoch": 0.359168997203356,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.1139,
      "step": 899
    },
    {
      "epoch": 0.3595685177786656,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.1054,
      "step": 900
    },
    {
      "epoch": 0.35996803835397523,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.1211,
      "step": 901
    },
    {
      "epoch": 0.36036755892928485,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.1226,
      "step": 902
    },
    {
      "epoch": 0.36076707950459447,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.1124,
      "step": 903
    },
    {
      "epoch": 0.3611666000799041,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.1111,
      "step": 904
    },
    {
      "epoch": 0.36156612065521376,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.1186,
      "step": 905
    },
    {
      "epoch": 0.3619656412305234,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.1333,
      "step": 906
    },
    {
      "epoch": 0.362365161805833,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.1039,
      "step": 907
    },
    {
      "epoch": 0.3627646823811426,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.111,
      "step": 908
    },
    {
      "epoch": 0.36316420295645224,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.1,
      "step": 909
    },
    {
      "epoch": 0.36356372353176186,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.12,
      "step": 910
    },
    {
      "epoch": 0.36396324410707154,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.1181,
      "step": 911
    },
    {
      "epoch": 0.36436276468238116,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.1208,
      "step": 912
    },
    {
      "epoch": 0.3647622852576908,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.1053,
      "step": 913
    },
    {
      "epoch": 0.3651618058330004,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.1272,
      "step": 914
    },
    {
      "epoch": 0.36556132640831,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.1026,
      "step": 915
    },
    {
      "epoch": 0.36596084698361964,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.1091,
      "step": 916
    },
    {
      "epoch": 0.3663603675589293,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.1176,
      "step": 917
    },
    {
      "epoch": 0.36675988813423893,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.1093,
      "step": 918
    },
    {
      "epoch": 0.36715940870954855,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.1085,
      "step": 919
    },
    {
      "epoch": 0.36755892928485817,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.1154,
      "step": 920
    },
    {
      "epoch": 0.3679584498601678,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.1095,
      "step": 921
    },
    {
      "epoch": 0.3683579704354774,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.1023,
      "step": 922
    },
    {
      "epoch": 0.36875749101078703,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.117,
      "step": 923
    },
    {
      "epoch": 0.3691570115860967,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0005,
      "loss": 1.1042,
      "step": 924
    },
    {
      "epoch": 0.3695565321614063,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.1231,
      "step": 925
    },
    {
      "epoch": 0.36995605273671595,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.1149,
      "step": 926
    },
    {
      "epoch": 0.37035557331202557,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.109,
      "step": 927
    },
    {
      "epoch": 0.3707550938873352,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.1165,
      "step": 928
    },
    {
      "epoch": 0.3711546144626448,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.126,
      "step": 929
    },
    {
      "epoch": 0.3715541350379545,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.1044,
      "step": 930
    },
    {
      "epoch": 0.3719536556132641,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.112,
      "step": 931
    },
    {
      "epoch": 0.3723531761885737,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.1212,
      "step": 932
    },
    {
      "epoch": 0.37275269676388334,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.1152,
      "step": 933
    },
    {
      "epoch": 0.37315221733919296,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.1004,
      "step": 934
    },
    {
      "epoch": 0.3735517379145026,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.115,
      "step": 935
    },
    {
      "epoch": 0.3739512584898122,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.1087,
      "step": 936
    },
    {
      "epoch": 0.3743507790651219,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.1125,
      "step": 937
    },
    {
      "epoch": 0.3747502996404315,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.106,
      "step": 938
    },
    {
      "epoch": 0.3751498202157411,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.105,
      "step": 939
    },
    {
      "epoch": 0.37554934079105073,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.1088,
      "step": 940
    },
    {
      "epoch": 0.37594886136636035,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.1079,
      "step": 941
    },
    {
      "epoch": 0.37634838194167,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0005,
      "loss": 1.1259,
      "step": 942
    },
    {
      "epoch": 0.37674790251697965,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.1178,
      "step": 943
    },
    {
      "epoch": 0.37714742309228927,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.1171,
      "step": 944
    },
    {
      "epoch": 0.3775469436675989,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0005,
      "loss": 1.1141,
      "step": 945
    },
    {
      "epoch": 0.3779464642429085,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.1031,
      "step": 946
    },
    {
      "epoch": 0.37834598481821813,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.1032,
      "step": 947
    },
    {
      "epoch": 0.37874550539352775,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.1065,
      "step": 948
    },
    {
      "epoch": 0.37914502596883737,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.1026,
      "step": 949
    },
    {
      "epoch": 0.37954454654414704,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.1033,
      "step": 950
    },
    {
      "epoch": 0.37994406711945666,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0005,
      "loss": 1.1065,
      "step": 951
    },
    {
      "epoch": 0.3803435876947663,
      "grad_norm": 0.25,
      "learning_rate": 0.0005,
      "loss": 1.0993,
      "step": 952
    },
    {
      "epoch": 0.3807431082700759,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.1102,
      "step": 953
    },
    {
      "epoch": 0.3811426288453855,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005,
      "loss": 1.1112,
      "step": 954
    },
    {
      "epoch": 0.38154214942069514,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005,
      "loss": 1.1267,
      "step": 955
    },
    {
      "epoch": 0.3819416699960048,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.1136,
      "step": 956
    },
    {
      "epoch": 0.38234119057131444,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0005,
      "loss": 1.1153,
      "step": 957
    },
    {
      "epoch": 0.38274071114662406,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.1176,
      "step": 958
    },
    {
      "epoch": 0.3831402317219337,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.1032,
      "step": 959
    },
    {
      "epoch": 0.3835397522972433,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.098,
      "step": 960
    },
    {
      "epoch": 0.3839392728725529,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.1079,
      "step": 961
    },
    {
      "epoch": 0.3843387934478626,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.1118,
      "step": 962
    },
    {
      "epoch": 0.3847383140231722,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.11,
      "step": 963
    },
    {
      "epoch": 0.38513783459848183,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.114,
      "step": 964
    },
    {
      "epoch": 0.38553735517379145,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.1,
      "step": 965
    },
    {
      "epoch": 0.38593687574910107,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.1117,
      "step": 966
    },
    {
      "epoch": 0.3863363963244107,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.1025,
      "step": 967
    },
    {
      "epoch": 0.3867359168997203,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.105,
      "step": 968
    },
    {
      "epoch": 0.38713543747503,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0999,
      "step": 969
    },
    {
      "epoch": 0.3875349580503396,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.1003,
      "step": 970
    },
    {
      "epoch": 0.3879344786256492,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.1154,
      "step": 971
    },
    {
      "epoch": 0.38833399920095885,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.1136,
      "step": 972
    },
    {
      "epoch": 0.38873351977626847,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.1006,
      "step": 973
    },
    {
      "epoch": 0.3891330403515781,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0997,
      "step": 974
    },
    {
      "epoch": 0.38953256092688776,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.1163,
      "step": 975
    },
    {
      "epoch": 0.3899320815021974,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.1066,
      "step": 976
    },
    {
      "epoch": 0.390331602077507,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.1026,
      "step": 977
    },
    {
      "epoch": 0.3907311226528166,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.1129,
      "step": 978
    },
    {
      "epoch": 0.39113064322812624,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0005,
      "loss": 1.0989,
      "step": 979
    },
    {
      "epoch": 0.39153016380343586,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0005,
      "loss": 1.0997,
      "step": 980
    },
    {
      "epoch": 0.3919296843787455,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.1098,
      "step": 981
    },
    {
      "epoch": 0.39232920495405516,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.1177,
      "step": 982
    },
    {
      "epoch": 0.3927287255293648,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.1183,
      "step": 983
    },
    {
      "epoch": 0.3931282461046744,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.1243,
      "step": 984
    },
    {
      "epoch": 0.393527766679984,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.1094,
      "step": 985
    },
    {
      "epoch": 0.39392728725529363,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.1243,
      "step": 986
    },
    {
      "epoch": 0.39432680783060325,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.1172,
      "step": 987
    },
    {
      "epoch": 0.39472632840591293,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.1021,
      "step": 988
    },
    {
      "epoch": 0.39512584898122255,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.1192,
      "step": 989
    },
    {
      "epoch": 0.39552536955653217,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.11,
      "step": 990
    },
    {
      "epoch": 0.3959248901318418,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.1117,
      "step": 991
    },
    {
      "epoch": 0.3963244107071514,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.1064,
      "step": 992
    },
    {
      "epoch": 0.39672393128246103,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0968,
      "step": 993
    },
    {
      "epoch": 0.39712345185777065,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.1097,
      "step": 994
    },
    {
      "epoch": 0.3975229724330803,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.1062,
      "step": 995
    },
    {
      "epoch": 0.39792249300838994,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.1056,
      "step": 996
    },
    {
      "epoch": 0.39832201358369956,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.1188,
      "step": 997
    },
    {
      "epoch": 0.3987215341590092,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.1159,
      "step": 998
    },
    {
      "epoch": 0.3991210547343188,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.117,
      "step": 999
    },
    {
      "epoch": 0.3995205753096284,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.1096,
      "step": 1000
    },
    {
      "epoch": 0.3999200958849381,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0005,
      "loss": 1.0965,
      "step": 1001
    },
    {
      "epoch": 0.4003196164602477,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.1007,
      "step": 1002
    },
    {
      "epoch": 0.40071913703555734,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.1099,
      "step": 1003
    },
    {
      "epoch": 0.40111865761086696,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.11,
      "step": 1004
    },
    {
      "epoch": 0.4015181781861766,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.1132,
      "step": 1005
    },
    {
      "epoch": 0.4019176987614862,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.1055,
      "step": 1006
    },
    {
      "epoch": 0.4023172193367959,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.1181,
      "step": 1007
    },
    {
      "epoch": 0.4027167399121055,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.1068,
      "step": 1008
    },
    {
      "epoch": 0.4031162604874151,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.114,
      "step": 1009
    },
    {
      "epoch": 0.40351578106272473,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.1061,
      "step": 1010
    },
    {
      "epoch": 0.40391530163803435,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.1111,
      "step": 1011
    },
    {
      "epoch": 0.40431482221334397,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.1056,
      "step": 1012
    },
    {
      "epoch": 0.4047143427886536,
      "grad_norm": 0.265625,
      "learning_rate": 0.0005,
      "loss": 1.1122,
      "step": 1013
    },
    {
      "epoch": 0.40511386336396327,
      "grad_norm": 0.25,
      "learning_rate": 0.0005,
      "loss": 1.103,
      "step": 1014
    },
    {
      "epoch": 0.4055133839392729,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0837,
      "step": 1015
    },
    {
      "epoch": 0.4059129045145825,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0958,
      "step": 1016
    },
    {
      "epoch": 0.4063124250898921,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.113,
      "step": 1017
    },
    {
      "epoch": 0.40671194566520175,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.1144,
      "step": 1018
    },
    {
      "epoch": 0.40711146624051137,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.1026,
      "step": 1019
    },
    {
      "epoch": 0.40751098681582104,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.0997,
      "step": 1020
    },
    {
      "epoch": 0.40791050739113066,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0005,
      "loss": 1.1152,
      "step": 1021
    },
    {
      "epoch": 0.4083100279664403,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.1046,
      "step": 1022
    },
    {
      "epoch": 0.4087095485417499,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.1062,
      "step": 1023
    },
    {
      "epoch": 0.4091090691170595,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.1174,
      "step": 1024
    },
    {
      "epoch": 0.40950858969236914,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0985,
      "step": 1025
    },
    {
      "epoch": 0.40990811026767876,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.107,
      "step": 1026
    },
    {
      "epoch": 0.41030763084298844,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0986,
      "step": 1027
    },
    {
      "epoch": 0.41070715141829806,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.1137,
      "step": 1028
    },
    {
      "epoch": 0.4111066719936077,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.1065,
      "step": 1029
    },
    {
      "epoch": 0.4115061925689173,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0005,
      "loss": 1.1051,
      "step": 1030
    },
    {
      "epoch": 0.4119057131442269,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.0995,
      "step": 1031
    },
    {
      "epoch": 0.41230523371953653,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.1028,
      "step": 1032
    },
    {
      "epoch": 0.4127047542948462,
      "grad_norm": 0.296875,
      "learning_rate": 0.0005,
      "loss": 1.1069,
      "step": 1033
    },
    {
      "epoch": 0.41310427487015583,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0005,
      "loss": 1.1155,
      "step": 1034
    },
    {
      "epoch": 0.41350379544546545,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.1184,
      "step": 1035
    },
    {
      "epoch": 0.41390331602077507,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.1085,
      "step": 1036
    },
    {
      "epoch": 0.4143028365960847,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.1053,
      "step": 1037
    },
    {
      "epoch": 0.4147023571713943,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.111,
      "step": 1038
    },
    {
      "epoch": 0.41510187774670393,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.1093,
      "step": 1039
    },
    {
      "epoch": 0.4155013983220136,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0993,
      "step": 1040
    },
    {
      "epoch": 0.4159009188973232,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.1076,
      "step": 1041
    },
    {
      "epoch": 0.41630043947263284,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.105,
      "step": 1042
    },
    {
      "epoch": 0.41669996004794246,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.1077,
      "step": 1043
    },
    {
      "epoch": 0.4170994806232521,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.1102,
      "step": 1044
    },
    {
      "epoch": 0.4174990011985617,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.1026,
      "step": 1045
    },
    {
      "epoch": 0.4178985217738714,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.102,
      "step": 1046
    },
    {
      "epoch": 0.418298042349181,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.1051,
      "step": 1047
    },
    {
      "epoch": 0.4186975629244906,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0991,
      "step": 1048
    },
    {
      "epoch": 0.41909708349980024,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.1085,
      "step": 1049
    },
    {
      "epoch": 0.41949660407510986,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.107,
      "step": 1050
    },
    {
      "epoch": 0.4198961246504195,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.1038,
      "step": 1051
    },
    {
      "epoch": 0.4202956452257291,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0005,
      "loss": 1.1047,
      "step": 1052
    },
    {
      "epoch": 0.4206951658010388,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0005,
      "loss": 1.0949,
      "step": 1053
    },
    {
      "epoch": 0.4210946863763484,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.1119,
      "step": 1054
    },
    {
      "epoch": 0.421494206951658,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0005,
      "loss": 1.106,
      "step": 1055
    },
    {
      "epoch": 0.42189372752696763,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0005,
      "loss": 1.1006,
      "step": 1056
    },
    {
      "epoch": 0.42229324810227725,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.108,
      "step": 1057
    },
    {
      "epoch": 0.42269276867758687,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0005,
      "loss": 1.0981,
      "step": 1058
    },
    {
      "epoch": 0.42309228925289655,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005,
      "loss": 1.1095,
      "step": 1059
    },
    {
      "epoch": 0.42349180982820617,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.1015,
      "step": 1060
    },
    {
      "epoch": 0.4238913304035158,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005,
      "loss": 1.1184,
      "step": 1061
    },
    {
      "epoch": 0.4242908509788254,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0005,
      "loss": 1.0952,
      "step": 1062
    },
    {
      "epoch": 0.424690371554135,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.1005,
      "step": 1063
    },
    {
      "epoch": 0.42508989212944465,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005,
      "loss": 1.1064,
      "step": 1064
    },
    {
      "epoch": 0.4254894127047543,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0005,
      "loss": 1.106,
      "step": 1065
    },
    {
      "epoch": 0.42588893328006394,
      "grad_norm": 0.546875,
      "learning_rate": 0.0005,
      "loss": 1.0977,
      "step": 1066
    },
    {
      "epoch": 0.42628845385537356,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0005,
      "loss": 1.1119,
      "step": 1067
    },
    {
      "epoch": 0.4266879744306832,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0005,
      "loss": 1.1161,
      "step": 1068
    },
    {
      "epoch": 0.4270874950059928,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0005,
      "loss": 1.0958,
      "step": 1069
    },
    {
      "epoch": 0.4274870155813024,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0005,
      "loss": 1.1097,
      "step": 1070
    },
    {
      "epoch": 0.42788653615661204,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0005,
      "loss": 1.1184,
      "step": 1071
    },
    {
      "epoch": 0.4282860567319217,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0005,
      "loss": 1.1183,
      "step": 1072
    },
    {
      "epoch": 0.42868557730723134,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0005,
      "loss": 1.1076,
      "step": 1073
    },
    {
      "epoch": 0.42908509788254096,
      "grad_norm": 0.375,
      "learning_rate": 0.0005,
      "loss": 1.1098,
      "step": 1074
    },
    {
      "epoch": 0.4294846184578506,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0005,
      "loss": 1.1199,
      "step": 1075
    },
    {
      "epoch": 0.4298841390331602,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0005,
      "loss": 1.1058,
      "step": 1076
    },
    {
      "epoch": 0.4302836596084698,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0005,
      "loss": 1.1095,
      "step": 1077
    },
    {
      "epoch": 0.4306831801837795,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0005,
      "loss": 1.1184,
      "step": 1078
    },
    {
      "epoch": 0.4310827007590891,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0886,
      "step": 1079
    },
    {
      "epoch": 0.43148222133439873,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005,
      "loss": 1.111,
      "step": 1080
    },
    {
      "epoch": 0.43188174190970835,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005,
      "loss": 1.0993,
      "step": 1081
    },
    {
      "epoch": 0.43228126248501797,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0005,
      "loss": 1.1068,
      "step": 1082
    },
    {
      "epoch": 0.4326807830603276,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0005,
      "loss": 1.1311,
      "step": 1083
    },
    {
      "epoch": 0.4330803036356372,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0005,
      "loss": 1.0983,
      "step": 1084
    },
    {
      "epoch": 0.4334798242109469,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0005,
      "loss": 1.1168,
      "step": 1085
    },
    {
      "epoch": 0.4338793447862565,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0005,
      "loss": 1.1093,
      "step": 1086
    },
    {
      "epoch": 0.4342788653615661,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0005,
      "loss": 1.1235,
      "step": 1087
    },
    {
      "epoch": 0.43467838593687574,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005,
      "loss": 1.1045,
      "step": 1088
    },
    {
      "epoch": 0.43507790651218536,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.1013,
      "step": 1089
    },
    {
      "epoch": 0.435477427087495,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0005,
      "loss": 1.1054,
      "step": 1090
    },
    {
      "epoch": 0.43587694766280466,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.1,
      "step": 1091
    },
    {
      "epoch": 0.4362764682381143,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0005,
      "loss": 1.1086,
      "step": 1092
    },
    {
      "epoch": 0.4366759888134239,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.1044,
      "step": 1093
    },
    {
      "epoch": 0.4370755093887335,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.1061,
      "step": 1094
    },
    {
      "epoch": 0.43747502996404314,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.1059,
      "step": 1095
    },
    {
      "epoch": 0.43787455053935276,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.109,
      "step": 1096
    },
    {
      "epoch": 0.4382740711146624,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.1059,
      "step": 1097
    },
    {
      "epoch": 0.43867359168997205,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.1094,
      "step": 1098
    },
    {
      "epoch": 0.4390731122652817,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0991,
      "step": 1099
    },
    {
      "epoch": 0.4394726328405913,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.1029,
      "step": 1100
    },
    {
      "epoch": 0.4398721534159009,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.1135,
      "step": 1101
    },
    {
      "epoch": 0.44027167399121053,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0877,
      "step": 1102
    },
    {
      "epoch": 0.44067119456652015,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.1105,
      "step": 1103
    },
    {
      "epoch": 0.4410707151418298,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0954,
      "step": 1104
    },
    {
      "epoch": 0.44147023571713945,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0878,
      "step": 1105
    },
    {
      "epoch": 0.44186975629244907,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0994,
      "step": 1106
    },
    {
      "epoch": 0.4422692768677587,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0992,
      "step": 1107
    },
    {
      "epoch": 0.4426687974430683,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.1084,
      "step": 1108
    },
    {
      "epoch": 0.4430683180183779,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0918,
      "step": 1109
    },
    {
      "epoch": 0.4434678385936876,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.1117,
      "step": 1110
    },
    {
      "epoch": 0.4438673591689972,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.097,
      "step": 1111
    },
    {
      "epoch": 0.44426687974430684,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0944,
      "step": 1112
    },
    {
      "epoch": 0.44466640031961646,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.1118,
      "step": 1113
    },
    {
      "epoch": 0.4450659208949261,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.1046,
      "step": 1114
    },
    {
      "epoch": 0.4454654414702357,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.107,
      "step": 1115
    },
    {
      "epoch": 0.4458649620455453,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0954,
      "step": 1116
    },
    {
      "epoch": 0.446264482620855,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.1131,
      "step": 1117
    },
    {
      "epoch": 0.4466640031961646,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0977,
      "step": 1118
    },
    {
      "epoch": 0.44706352377147424,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.094,
      "step": 1119
    },
    {
      "epoch": 0.44746304434678386,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0968,
      "step": 1120
    },
    {
      "epoch": 0.4478625649220935,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0005,
      "loss": 1.1049,
      "step": 1121
    },
    {
      "epoch": 0.4482620854974031,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.1015,
      "step": 1122
    },
    {
      "epoch": 0.44866160607271277,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0005,
      "loss": 1.1017,
      "step": 1123
    },
    {
      "epoch": 0.4490611266480224,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0005,
      "loss": 1.1021,
      "step": 1124
    },
    {
      "epoch": 0.449460647223332,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.1097,
      "step": 1125
    },
    {
      "epoch": 0.44986016779864163,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.1063,
      "step": 1126
    },
    {
      "epoch": 0.45025968837395125,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.1025,
      "step": 1127
    },
    {
      "epoch": 0.45065920894926087,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0005,
      "loss": 1.1174,
      "step": 1128
    },
    {
      "epoch": 0.4510587295245705,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0986,
      "step": 1129
    },
    {
      "epoch": 0.45145825009988017,
      "grad_norm": 0.203125,
      "learning_rate": 0.0005,
      "loss": 1.1,
      "step": 1130
    },
    {
      "epoch": 0.4518577706751898,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0005,
      "loss": 1.1016,
      "step": 1131
    },
    {
      "epoch": 0.4522572912504994,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0972,
      "step": 1132
    },
    {
      "epoch": 0.452656811825809,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.1077,
      "step": 1133
    },
    {
      "epoch": 0.45305633240111864,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.1007,
      "step": 1134
    },
    {
      "epoch": 0.45345585297642826,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.1082,
      "step": 1135
    },
    {
      "epoch": 0.45385537355173794,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0882,
      "step": 1136
    },
    {
      "epoch": 0.45425489412704756,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.1097,
      "step": 1137
    },
    {
      "epoch": 0.4546544147023572,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0933,
      "step": 1138
    },
    {
      "epoch": 0.4550539352776668,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.1135,
      "step": 1139
    },
    {
      "epoch": 0.4554534558529764,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.1041,
      "step": 1140
    },
    {
      "epoch": 0.45585297642828604,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.1054,
      "step": 1141
    },
    {
      "epoch": 0.45625249700359566,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.1036,
      "step": 1142
    },
    {
      "epoch": 0.45665201757890533,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.1019,
      "step": 1143
    },
    {
      "epoch": 0.45705153815421495,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.1012,
      "step": 1144
    },
    {
      "epoch": 0.4574510587295246,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.096,
      "step": 1145
    },
    {
      "epoch": 0.4578505793048342,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0997,
      "step": 1146
    },
    {
      "epoch": 0.4582500998801438,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0976,
      "step": 1147
    },
    {
      "epoch": 0.45864962045545343,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.1063,
      "step": 1148
    },
    {
      "epoch": 0.4590491410307631,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0968,
      "step": 1149
    },
    {
      "epoch": 0.45944866160607273,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0933,
      "step": 1150
    },
    {
      "epoch": 0.45984818218138235,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.1062,
      "step": 1151
    },
    {
      "epoch": 0.46024770275669197,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.1137,
      "step": 1152
    },
    {
      "epoch": 0.4606472233320016,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.1079,
      "step": 1153
    },
    {
      "epoch": 0.4610467439073112,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0865,
      "step": 1154
    },
    {
      "epoch": 0.4614462644826209,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.1008,
      "step": 1155
    },
    {
      "epoch": 0.4618457850579305,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.1076,
      "step": 1156
    },
    {
      "epoch": 0.4622453056332401,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.1033,
      "step": 1157
    },
    {
      "epoch": 0.46264482620854974,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.1003,
      "step": 1158
    },
    {
      "epoch": 0.46304434678385936,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.1013,
      "step": 1159
    },
    {
      "epoch": 0.463443867359169,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0869,
      "step": 1160
    },
    {
      "epoch": 0.4638433879344786,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0962,
      "step": 1161
    },
    {
      "epoch": 0.4642429085097883,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0988,
      "step": 1162
    },
    {
      "epoch": 0.4646424290850979,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0962,
      "step": 1163
    },
    {
      "epoch": 0.4650419496604075,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.1197,
      "step": 1164
    },
    {
      "epoch": 0.46544147023571714,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.1057,
      "step": 1165
    },
    {
      "epoch": 0.46584099081102676,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.114,
      "step": 1166
    },
    {
      "epoch": 0.4662405113863364,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.1031,
      "step": 1167
    },
    {
      "epoch": 0.46664003196164605,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0964,
      "step": 1168
    },
    {
      "epoch": 0.46703955253695567,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.1088,
      "step": 1169
    },
    {
      "epoch": 0.4674390731122653,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.097,
      "step": 1170
    },
    {
      "epoch": 0.4678385936875749,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.1086,
      "step": 1171
    },
    {
      "epoch": 0.46823811426288453,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0996,
      "step": 1172
    },
    {
      "epoch": 0.46863763483819415,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0955,
      "step": 1173
    },
    {
      "epoch": 0.46903715541350377,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.1042,
      "step": 1174
    },
    {
      "epoch": 0.46943667598881345,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.1077,
      "step": 1175
    },
    {
      "epoch": 0.46983619656412307,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0956,
      "step": 1176
    },
    {
      "epoch": 0.4702357171394327,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.1061,
      "step": 1177
    },
    {
      "epoch": 0.4706352377147423,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.1025,
      "step": 1178
    },
    {
      "epoch": 0.4710347582900519,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.1008,
      "step": 1179
    },
    {
      "epoch": 0.47143427886536154,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0991,
      "step": 1180
    },
    {
      "epoch": 0.4718337994406712,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.1065,
      "step": 1181
    },
    {
      "epoch": 0.47223332001598084,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0864,
      "step": 1182
    },
    {
      "epoch": 0.47263284059129046,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.1035,
      "step": 1183
    },
    {
      "epoch": 0.4730323611666001,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.1053,
      "step": 1184
    },
    {
      "epoch": 0.4734318817419097,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.1088,
      "step": 1185
    },
    {
      "epoch": 0.4738314023172193,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0987,
      "step": 1186
    },
    {
      "epoch": 0.47423092289252894,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0884,
      "step": 1187
    },
    {
      "epoch": 0.4746304434678386,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.1026,
      "step": 1188
    },
    {
      "epoch": 0.47502996404314823,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0953,
      "step": 1189
    },
    {
      "epoch": 0.47542948461845785,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.1048,
      "step": 1190
    },
    {
      "epoch": 0.4758290051937675,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.1037,
      "step": 1191
    },
    {
      "epoch": 0.4762285257690771,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.1056,
      "step": 1192
    },
    {
      "epoch": 0.4766280463443867,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0974,
      "step": 1193
    },
    {
      "epoch": 0.4770275669196964,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0984,
      "step": 1194
    },
    {
      "epoch": 0.477427087495006,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.081,
      "step": 1195
    },
    {
      "epoch": 0.47782660807031563,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.106,
      "step": 1196
    },
    {
      "epoch": 0.47822612864562525,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.1106,
      "step": 1197
    },
    {
      "epoch": 0.47862564922093487,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.1059,
      "step": 1198
    },
    {
      "epoch": 0.4790251697962445,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0934,
      "step": 1199
    },
    {
      "epoch": 0.4794246903715541,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.1034,
      "step": 1200
    },
    {
      "epoch": 0.4798242109468638,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.1015,
      "step": 1201
    },
    {
      "epoch": 0.4802237315221734,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.1113,
      "step": 1202
    },
    {
      "epoch": 0.480623252097483,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.1088,
      "step": 1203
    },
    {
      "epoch": 0.48102277267279264,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.1086,
      "step": 1204
    },
    {
      "epoch": 0.48142229324810226,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.1095,
      "step": 1205
    },
    {
      "epoch": 0.4818218138234119,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0936,
      "step": 1206
    },
    {
      "epoch": 0.48222133439872156,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0981,
      "step": 1207
    },
    {
      "epoch": 0.4826208549740312,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0949,
      "step": 1208
    },
    {
      "epoch": 0.4830203755493408,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.1047,
      "step": 1209
    },
    {
      "epoch": 0.4834198961246504,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.1044,
      "step": 1210
    },
    {
      "epoch": 0.48381941669996004,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0869,
      "step": 1211
    },
    {
      "epoch": 0.48421893727526966,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.1052,
      "step": 1212
    },
    {
      "epoch": 0.48461845785057933,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.093,
      "step": 1213
    },
    {
      "epoch": 0.48501797842588895,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0988,
      "step": 1214
    },
    {
      "epoch": 0.48541749900119857,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.1106,
      "step": 1215
    },
    {
      "epoch": 0.4858170195765082,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0005,
      "loss": 1.0976,
      "step": 1216
    },
    {
      "epoch": 0.4862165401518178,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.1056,
      "step": 1217
    },
    {
      "epoch": 0.48661606072712743,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.1132,
      "step": 1218
    },
    {
      "epoch": 0.48701558130243705,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0918,
      "step": 1219
    },
    {
      "epoch": 0.4874151018777467,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.1038,
      "step": 1220
    },
    {
      "epoch": 0.48781462245305635,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0982,
      "step": 1221
    },
    {
      "epoch": 0.48821414302836597,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.1068,
      "step": 1222
    },
    {
      "epoch": 0.4886136636036756,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.1056,
      "step": 1223
    },
    {
      "epoch": 0.4890131841789852,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.1067,
      "step": 1224
    },
    {
      "epoch": 0.4894127047542948,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.1066,
      "step": 1225
    },
    {
      "epoch": 0.4898122253296045,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0924,
      "step": 1226
    },
    {
      "epoch": 0.4902117459049141,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0892,
      "step": 1227
    },
    {
      "epoch": 0.49061126648022374,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0965,
      "step": 1228
    },
    {
      "epoch": 0.49101078705553336,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.098,
      "step": 1229
    },
    {
      "epoch": 0.491410307630843,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.1095,
      "step": 1230
    },
    {
      "epoch": 0.4918098282061526,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.1062,
      "step": 1231
    },
    {
      "epoch": 0.4922093487814622,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.1001,
      "step": 1232
    },
    {
      "epoch": 0.4926088693567719,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0863,
      "step": 1233
    },
    {
      "epoch": 0.4930083899320815,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.1053,
      "step": 1234
    },
    {
      "epoch": 0.49340791050739113,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0979,
      "step": 1235
    },
    {
      "epoch": 0.49380743108270075,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0937,
      "step": 1236
    },
    {
      "epoch": 0.4942069516580104,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.1113,
      "step": 1237
    },
    {
      "epoch": 0.49460647223332,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0898,
      "step": 1238
    },
    {
      "epoch": 0.49500599280862967,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0914,
      "step": 1239
    },
    {
      "epoch": 0.4954055133839393,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0871,
      "step": 1240
    },
    {
      "epoch": 0.4958050339592489,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.1037,
      "step": 1241
    },
    {
      "epoch": 0.49620455453455853,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0938,
      "step": 1242
    },
    {
      "epoch": 0.49660407510986815,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0929,
      "step": 1243
    },
    {
      "epoch": 0.49700359568517777,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0903,
      "step": 1244
    },
    {
      "epoch": 0.4974031162604874,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.1018,
      "step": 1245
    },
    {
      "epoch": 0.49780263683579706,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.1063,
      "step": 1246
    },
    {
      "epoch": 0.4982021574111067,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.1045,
      "step": 1247
    },
    {
      "epoch": 0.4986016779864163,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0994,
      "step": 1248
    },
    {
      "epoch": 0.4990011985617259,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0974,
      "step": 1249
    },
    {
      "epoch": 0.49940071913703554,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0989,
      "step": 1250
    },
    {
      "epoch": 0.49980023971234516,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.1052,
      "step": 1251
    },
    {
      "epoch": 0.5001997602876548,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0005,
      "loss": 1.1056,
      "step": 1252
    },
    {
      "epoch": 0.5005992808629645,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0005,
      "loss": 1.0985,
      "step": 1253
    },
    {
      "epoch": 0.500998801438274,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0951,
      "step": 1254
    },
    {
      "epoch": 0.5013983220135837,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.1023,
      "step": 1255
    },
    {
      "epoch": 0.5017978425888934,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0969,
      "step": 1256
    },
    {
      "epoch": 0.5021973631642029,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0939,
      "step": 1257
    },
    {
      "epoch": 0.5025968837395126,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0974,
      "step": 1258
    },
    {
      "epoch": 0.5029964043148222,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0887,
      "step": 1259
    },
    {
      "epoch": 0.5033959248901319,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.1099,
      "step": 1260
    },
    {
      "epoch": 0.5037954454654414,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.1097,
      "step": 1261
    },
    {
      "epoch": 0.5041949660407511,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.1036,
      "step": 1262
    },
    {
      "epoch": 0.5045944866160608,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.1126,
      "step": 1263
    },
    {
      "epoch": 0.5049940071913703,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0953,
      "step": 1264
    },
    {
      "epoch": 0.50539352776668,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0883,
      "step": 1265
    },
    {
      "epoch": 0.5057930483419896,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0981,
      "step": 1266
    },
    {
      "epoch": 0.5061925689172992,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.1096,
      "step": 1267
    },
    {
      "epoch": 0.5065920894926089,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.1091,
      "step": 1268
    },
    {
      "epoch": 0.5069916100679185,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0957,
      "step": 1269
    },
    {
      "epoch": 0.5073911306432282,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.1023,
      "step": 1270
    },
    {
      "epoch": 0.5077906512185377,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.1057,
      "step": 1271
    },
    {
      "epoch": 0.5081901717938474,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0922,
      "step": 1272
    },
    {
      "epoch": 0.508589692369157,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0869,
      "step": 1273
    },
    {
      "epoch": 0.5089892129444666,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0994,
      "step": 1274
    },
    {
      "epoch": 0.5093887335197763,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0919,
      "step": 1275
    },
    {
      "epoch": 0.5097882540950859,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0908,
      "step": 1276
    },
    {
      "epoch": 0.5101877746703956,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0005,
      "loss": 1.107,
      "step": 1277
    },
    {
      "epoch": 0.5105872952457051,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.1076,
      "step": 1278
    },
    {
      "epoch": 0.5109868158210148,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0939,
      "step": 1279
    },
    {
      "epoch": 0.5113863363963244,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005,
      "loss": 1.0959,
      "step": 1280
    },
    {
      "epoch": 0.511785856971634,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.1068,
      "step": 1281
    },
    {
      "epoch": 0.5121853775469437,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0005,
      "loss": 1.0945,
      "step": 1282
    },
    {
      "epoch": 0.5125848981222533,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0005,
      "loss": 1.0891,
      "step": 1283
    },
    {
      "epoch": 0.512984418697563,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.1016,
      "step": 1284
    },
    {
      "epoch": 0.5133839392728725,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0005,
      "loss": 1.1033,
      "step": 1285
    },
    {
      "epoch": 0.5137834598481822,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0005,
      "loss": 1.1044,
      "step": 1286
    },
    {
      "epoch": 0.5141829804234919,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.1069,
      "step": 1287
    },
    {
      "epoch": 0.5145825009988014,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0928,
      "step": 1288
    },
    {
      "epoch": 0.5149820215741111,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0955,
      "step": 1289
    },
    {
      "epoch": 0.5153815421494207,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0005,
      "loss": 1.0982,
      "step": 1290
    },
    {
      "epoch": 0.5157810627247303,
      "grad_norm": 0.375,
      "learning_rate": 0.0005,
      "loss": 1.0972,
      "step": 1291
    },
    {
      "epoch": 0.5161805833000399,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005,
      "loss": 1.1148,
      "step": 1292
    },
    {
      "epoch": 0.5165801038753496,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.1026,
      "step": 1293
    },
    {
      "epoch": 0.5169796244506593,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0005,
      "loss": 1.0958,
      "step": 1294
    },
    {
      "epoch": 0.5173791450259688,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0005,
      "loss": 1.0981,
      "step": 1295
    },
    {
      "epoch": 0.5177786656012785,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0979,
      "step": 1296
    },
    {
      "epoch": 0.5181781861765881,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0005,
      "loss": 1.0915,
      "step": 1297
    },
    {
      "epoch": 0.5185777067518977,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0005,
      "loss": 1.0964,
      "step": 1298
    },
    {
      "epoch": 0.5189772273272073,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0994,
      "step": 1299
    },
    {
      "epoch": 0.519376747902517,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0005,
      "loss": 1.1063,
      "step": 1300
    },
    {
      "epoch": 0.5197762684778267,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.097,
      "step": 1301
    },
    {
      "epoch": 0.5201757890531362,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0005,
      "loss": 1.11,
      "step": 1302
    },
    {
      "epoch": 0.5205753096284459,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0948,
      "step": 1303
    },
    {
      "epoch": 0.5209748302037555,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0973,
      "step": 1304
    },
    {
      "epoch": 0.5213743507790651,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.1017,
      "step": 1305
    },
    {
      "epoch": 0.5217738713543747,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0796,
      "step": 1306
    },
    {
      "epoch": 0.5221733919296844,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0005,
      "loss": 1.089,
      "step": 1307
    },
    {
      "epoch": 0.522572912504994,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0005,
      "loss": 1.0997,
      "step": 1308
    },
    {
      "epoch": 0.5229724330803036,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.105,
      "step": 1309
    },
    {
      "epoch": 0.5233719536556133,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0005,
      "loss": 1.0892,
      "step": 1310
    },
    {
      "epoch": 0.5237714742309229,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.0884,
      "step": 1311
    },
    {
      "epoch": 0.5241709948062325,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0932,
      "step": 1312
    },
    {
      "epoch": 0.5245705153815422,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0005,
      "loss": 1.1035,
      "step": 1313
    },
    {
      "epoch": 0.5249700359568518,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0912,
      "step": 1314
    },
    {
      "epoch": 0.5253695565321614,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0998,
      "step": 1315
    },
    {
      "epoch": 0.525769077107471,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0954,
      "step": 1316
    },
    {
      "epoch": 0.5261685976827807,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.1071,
      "step": 1317
    },
    {
      "epoch": 0.5265681182580902,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0974,
      "step": 1318
    },
    {
      "epoch": 0.5269676388333999,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.1193,
      "step": 1319
    },
    {
      "epoch": 0.5273671594087096,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0956,
      "step": 1320
    },
    {
      "epoch": 0.5277666799840192,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.1009,
      "step": 1321
    },
    {
      "epoch": 0.5281662005593288,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0909,
      "step": 1322
    },
    {
      "epoch": 0.5285657211346384,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.1031,
      "step": 1323
    },
    {
      "epoch": 0.5289652417099481,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.1172,
      "step": 1324
    },
    {
      "epoch": 0.5293647622852576,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.1024,
      "step": 1325
    },
    {
      "epoch": 0.5297642828605673,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0851,
      "step": 1326
    },
    {
      "epoch": 0.530163803435877,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0951,
      "step": 1327
    },
    {
      "epoch": 0.5305633240111866,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0996,
      "step": 1328
    },
    {
      "epoch": 0.5309628445864962,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.1007,
      "step": 1329
    },
    {
      "epoch": 0.5313623651618058,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0997,
      "step": 1330
    },
    {
      "epoch": 0.5317618857371155,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0856,
      "step": 1331
    },
    {
      "epoch": 0.5321614063124251,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0947,
      "step": 1332
    },
    {
      "epoch": 0.5325609268877347,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0971,
      "step": 1333
    },
    {
      "epoch": 0.5329604474630444,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0949,
      "step": 1334
    },
    {
      "epoch": 0.533359968038354,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0992,
      "step": 1335
    },
    {
      "epoch": 0.5337594886136636,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0905,
      "step": 1336
    },
    {
      "epoch": 0.5341590091889732,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0944,
      "step": 1337
    },
    {
      "epoch": 0.5345585297642829,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0816,
      "step": 1338
    },
    {
      "epoch": 0.5349580503395925,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0881,
      "step": 1339
    },
    {
      "epoch": 0.5353575709149021,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0985,
      "step": 1340
    },
    {
      "epoch": 0.5357570914902118,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.1024,
      "step": 1341
    },
    {
      "epoch": 0.5361566120655213,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0909,
      "step": 1342
    },
    {
      "epoch": 0.536556132640831,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0936,
      "step": 1343
    },
    {
      "epoch": 0.5369556532161406,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.1062,
      "step": 1344
    },
    {
      "epoch": 0.5373551737914503,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0869,
      "step": 1345
    },
    {
      "epoch": 0.5377546943667599,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.093,
      "step": 1346
    },
    {
      "epoch": 0.5381542149420695,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0933,
      "step": 1347
    },
    {
      "epoch": 0.5385537355173792,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0921,
      "step": 1348
    },
    {
      "epoch": 0.5389532560926887,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0962,
      "step": 1349
    },
    {
      "epoch": 0.5393527766679984,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0991,
      "step": 1350
    },
    {
      "epoch": 0.539752297243308,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0964,
      "step": 1351
    },
    {
      "epoch": 0.5401518178186177,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0975,
      "step": 1352
    },
    {
      "epoch": 0.5405513383939273,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.1128,
      "step": 1353
    },
    {
      "epoch": 0.5409508589692369,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.114,
      "step": 1354
    },
    {
      "epoch": 0.5413503795445466,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.1042,
      "step": 1355
    },
    {
      "epoch": 0.5417499001198561,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0937,
      "step": 1356
    },
    {
      "epoch": 0.5421494206951658,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.0948,
      "step": 1357
    },
    {
      "epoch": 0.5425489412704755,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.1038,
      "step": 1358
    },
    {
      "epoch": 0.542948461845785,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.1042,
      "step": 1359
    },
    {
      "epoch": 0.5433479824210947,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.1009,
      "step": 1360
    },
    {
      "epoch": 0.5437475029964043,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.1107,
      "step": 1361
    },
    {
      "epoch": 0.544147023571714,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0923,
      "step": 1362
    },
    {
      "epoch": 0.5445465441470235,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.1038,
      "step": 1363
    },
    {
      "epoch": 0.5449460647223332,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.1006,
      "step": 1364
    },
    {
      "epoch": 0.5453455852976429,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0956,
      "step": 1365
    },
    {
      "epoch": 0.5457451058729524,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.099,
      "step": 1366
    },
    {
      "epoch": 0.5461446264482621,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.1045,
      "step": 1367
    },
    {
      "epoch": 0.5465441470235717,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0982,
      "step": 1368
    },
    {
      "epoch": 0.5469436675988814,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.1018,
      "step": 1369
    },
    {
      "epoch": 0.5473431881741909,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.1028,
      "step": 1370
    },
    {
      "epoch": 0.5477427087495006,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0954,
      "step": 1371
    },
    {
      "epoch": 0.5481422293248103,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0924,
      "step": 1372
    },
    {
      "epoch": 0.5485417499001198,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.0987,
      "step": 1373
    },
    {
      "epoch": 0.5489412704754295,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.1,
      "step": 1374
    },
    {
      "epoch": 0.5493407910507391,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0986,
      "step": 1375
    },
    {
      "epoch": 0.5497403116260487,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.1041,
      "step": 1376
    },
    {
      "epoch": 0.5501398322013584,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0957,
      "step": 1377
    },
    {
      "epoch": 0.550539352776668,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0926,
      "step": 1378
    },
    {
      "epoch": 0.5509388733519777,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.1007,
      "step": 1379
    },
    {
      "epoch": 0.5513383939272872,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0925,
      "step": 1380
    },
    {
      "epoch": 0.5517379145025969,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.1093,
      "step": 1381
    },
    {
      "epoch": 0.5521374350779065,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.1018,
      "step": 1382
    },
    {
      "epoch": 0.5525369556532161,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0938,
      "step": 1383
    },
    {
      "epoch": 0.5529364762285258,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0876,
      "step": 1384
    },
    {
      "epoch": 0.5533359968038354,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0922,
      "step": 1385
    },
    {
      "epoch": 0.5537355173791451,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0961,
      "step": 1386
    },
    {
      "epoch": 0.5541350379544546,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.1078,
      "step": 1387
    },
    {
      "epoch": 0.5545345585297643,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0975,
      "step": 1388
    },
    {
      "epoch": 0.5549340791050739,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0985,
      "step": 1389
    },
    {
      "epoch": 0.5553335996803835,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0978,
      "step": 1390
    },
    {
      "epoch": 0.5557331202556932,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0855,
      "step": 1391
    },
    {
      "epoch": 0.5561326408310028,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.097,
      "step": 1392
    },
    {
      "epoch": 0.5565321614063125,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.1086,
      "step": 1393
    },
    {
      "epoch": 0.556931681981622,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.1041,
      "step": 1394
    },
    {
      "epoch": 0.5573312025569317,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0905,
      "step": 1395
    },
    {
      "epoch": 0.5577307231322413,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0953,
      "step": 1396
    },
    {
      "epoch": 0.5581302437075509,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0983,
      "step": 1397
    },
    {
      "epoch": 0.5585297642828606,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.1132,
      "step": 1398
    },
    {
      "epoch": 0.5589292848581702,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0908,
      "step": 1399
    },
    {
      "epoch": 0.5593288054334798,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0939,
      "step": 1400
    },
    {
      "epoch": 0.5597283260087894,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0949,
      "step": 1401
    },
    {
      "epoch": 0.5601278465840991,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0894,
      "step": 1402
    },
    {
      "epoch": 0.5605273671594088,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0944,
      "step": 1403
    },
    {
      "epoch": 0.5609268877347183,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0912,
      "step": 1404
    },
    {
      "epoch": 0.561326408310028,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0865,
      "step": 1405
    },
    {
      "epoch": 0.5617259288853376,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.104,
      "step": 1406
    },
    {
      "epoch": 0.5621254494606472,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.097,
      "step": 1407
    },
    {
      "epoch": 0.5625249700359568,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.108,
      "step": 1408
    },
    {
      "epoch": 0.5629244906112665,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0967,
      "step": 1409
    },
    {
      "epoch": 0.5633240111865762,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0005,
      "loss": 1.0898,
      "step": 1410
    },
    {
      "epoch": 0.5637235317618857,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0904,
      "step": 1411
    },
    {
      "epoch": 0.5641230523371954,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.1134,
      "step": 1412
    },
    {
      "epoch": 0.564522572912505,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0888,
      "step": 1413
    },
    {
      "epoch": 0.5649220934878146,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0949,
      "step": 1414
    },
    {
      "epoch": 0.5653216140631242,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.1022,
      "step": 1415
    },
    {
      "epoch": 0.5657211346384339,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0896,
      "step": 1416
    },
    {
      "epoch": 0.5661206552137436,
      "grad_norm": 0.28125,
      "learning_rate": 0.0005,
      "loss": 1.0901,
      "step": 1417
    },
    {
      "epoch": 0.5665201757890531,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0005,
      "loss": 1.0901,
      "step": 1418
    },
    {
      "epoch": 0.5669196963643628,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0005,
      "loss": 1.1034,
      "step": 1419
    },
    {
      "epoch": 0.5673192169396724,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0945,
      "step": 1420
    },
    {
      "epoch": 0.567718737514982,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0005,
      "loss": 1.0945,
      "step": 1421
    },
    {
      "epoch": 0.5681182580902916,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.1028,
      "step": 1422
    },
    {
      "epoch": 0.5685177786656013,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0005,
      "loss": 1.0961,
      "step": 1423
    },
    {
      "epoch": 0.568917299240911,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0005,
      "loss": 1.0867,
      "step": 1424
    },
    {
      "epoch": 0.5693168198162205,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0933,
      "step": 1425
    },
    {
      "epoch": 0.5697163403915302,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0005,
      "loss": 1.1003,
      "step": 1426
    },
    {
      "epoch": 0.5701158609668397,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005,
      "loss": 1.0943,
      "step": 1427
    },
    {
      "epoch": 0.5705153815421494,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0005,
      "loss": 1.0856,
      "step": 1428
    },
    {
      "epoch": 0.5709149021174591,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0999,
      "step": 1429
    },
    {
      "epoch": 0.5713144226927687,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0005,
      "loss": 1.1085,
      "step": 1430
    },
    {
      "epoch": 0.5717139432680783,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0005,
      "loss": 1.0904,
      "step": 1431
    },
    {
      "epoch": 0.5721134638433879,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0943,
      "step": 1432
    },
    {
      "epoch": 0.5725129844186976,
      "grad_norm": 0.375,
      "learning_rate": 0.0005,
      "loss": 1.0827,
      "step": 1433
    },
    {
      "epoch": 0.5729125049940071,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0005,
      "loss": 1.0914,
      "step": 1434
    },
    {
      "epoch": 0.5733120255693168,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0005,
      "loss": 1.1115,
      "step": 1435
    },
    {
      "epoch": 0.5737115461446265,
      "grad_norm": 0.515625,
      "learning_rate": 0.0005,
      "loss": 1.1065,
      "step": 1436
    },
    {
      "epoch": 0.5741110667199361,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0005,
      "loss": 1.0973,
      "step": 1437
    },
    {
      "epoch": 0.5745105872952457,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0005,
      "loss": 1.0921,
      "step": 1438
    },
    {
      "epoch": 0.5749101078705553,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0005,
      "loss": 1.1002,
      "step": 1439
    },
    {
      "epoch": 0.575309628445865,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0874,
      "step": 1440
    },
    {
      "epoch": 0.5757091490211745,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0005,
      "loss": 1.0985,
      "step": 1441
    },
    {
      "epoch": 0.5761086695964842,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0877,
      "step": 1442
    },
    {
      "epoch": 0.5765081901717939,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.115,
      "step": 1443
    },
    {
      "epoch": 0.5769077107471035,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0922,
      "step": 1444
    },
    {
      "epoch": 0.5773072313224131,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.0783,
      "step": 1445
    },
    {
      "epoch": 0.5777067518977227,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0968,
      "step": 1446
    },
    {
      "epoch": 0.5781062724730324,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0005,
      "loss": 1.0875,
      "step": 1447
    },
    {
      "epoch": 0.578505793048342,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0933,
      "step": 1448
    },
    {
      "epoch": 0.5789053136236516,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.086,
      "step": 1449
    },
    {
      "epoch": 0.5793048341989613,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0908,
      "step": 1450
    },
    {
      "epoch": 0.5797043547742708,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.0888,
      "step": 1451
    },
    {
      "epoch": 0.5801038753495805,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.098,
      "step": 1452
    },
    {
      "epoch": 0.5805033959248901,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0975,
      "step": 1453
    },
    {
      "epoch": 0.5809029165001998,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.1069,
      "step": 1454
    },
    {
      "epoch": 0.5813024370755094,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0912,
      "step": 1455
    },
    {
      "epoch": 0.581701957650819,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0881,
      "step": 1456
    },
    {
      "epoch": 0.5821014782261287,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.087,
      "step": 1457
    },
    {
      "epoch": 0.5825009988014382,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0846,
      "step": 1458
    },
    {
      "epoch": 0.5829005193767479,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0902,
      "step": 1459
    },
    {
      "epoch": 0.5833000399520575,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0898,
      "step": 1460
    },
    {
      "epoch": 0.5836995605273672,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.1032,
      "step": 1461
    },
    {
      "epoch": 0.5840990811026768,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.1075,
      "step": 1462
    },
    {
      "epoch": 0.5844986016779864,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.1002,
      "step": 1463
    },
    {
      "epoch": 0.5848981222532961,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0987,
      "step": 1464
    },
    {
      "epoch": 0.5852976428286056,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0875,
      "step": 1465
    },
    {
      "epoch": 0.5856971634039153,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0942,
      "step": 1466
    },
    {
      "epoch": 0.5860966839792249,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0937,
      "step": 1467
    },
    {
      "epoch": 0.5864962045545346,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0974,
      "step": 1468
    },
    {
      "epoch": 0.5868957251298442,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0914,
      "step": 1469
    },
    {
      "epoch": 0.5872952457051538,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0822,
      "step": 1470
    },
    {
      "epoch": 0.5876947662804635,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0861,
      "step": 1471
    },
    {
      "epoch": 0.588094286855773,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.101,
      "step": 1472
    },
    {
      "epoch": 0.5884938074310827,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0898,
      "step": 1473
    },
    {
      "epoch": 0.5888933280063924,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0806,
      "step": 1474
    },
    {
      "epoch": 0.589292848581702,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0918,
      "step": 1475
    },
    {
      "epoch": 0.5896923691570116,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0909,
      "step": 1476
    },
    {
      "epoch": 0.5900918897323212,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0914,
      "step": 1477
    },
    {
      "epoch": 0.5904914103076309,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0954,
      "step": 1478
    },
    {
      "epoch": 0.5908909308829404,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0814,
      "step": 1479
    },
    {
      "epoch": 0.5912904514582501,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0868,
      "step": 1480
    },
    {
      "epoch": 0.5916899720335598,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0929,
      "step": 1481
    },
    {
      "epoch": 0.5920894926088693,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.102,
      "step": 1482
    },
    {
      "epoch": 0.592489013184179,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0926,
      "step": 1483
    },
    {
      "epoch": 0.5928885337594886,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.1009,
      "step": 1484
    },
    {
      "epoch": 0.5932880543347983,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0855,
      "step": 1485
    },
    {
      "epoch": 0.5936875749101078,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0804,
      "step": 1486
    },
    {
      "epoch": 0.5940870954854175,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0953,
      "step": 1487
    },
    {
      "epoch": 0.5944866160607272,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.1002,
      "step": 1488
    },
    {
      "epoch": 0.5948861366360367,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0845,
      "step": 1489
    },
    {
      "epoch": 0.5952856572113464,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0955,
      "step": 1490
    },
    {
      "epoch": 0.595685177786656,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0997,
      "step": 1491
    },
    {
      "epoch": 0.5960846983619656,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0999,
      "step": 1492
    },
    {
      "epoch": 0.5964842189372753,
      "grad_norm": 0.171875,
      "learning_rate": 0.0005,
      "loss": 1.0906,
      "step": 1493
    },
    {
      "epoch": 0.5968837395125849,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0884,
      "step": 1494
    },
    {
      "epoch": 0.5972832600878946,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0897,
      "step": 1495
    },
    {
      "epoch": 0.5976827806632041,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0005,
      "loss": 1.0903,
      "step": 1496
    },
    {
      "epoch": 0.5980823012385138,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0855,
      "step": 1497
    },
    {
      "epoch": 0.5984818218138234,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0005,
      "loss": 1.1,
      "step": 1498
    },
    {
      "epoch": 0.598881342389133,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0005,
      "loss": 1.0907,
      "step": 1499
    },
    {
      "epoch": 0.5992808629644427,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0945,
      "step": 1500
    },
    {
      "epoch": 0.5996803835397523,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.0956,
      "step": 1501
    },
    {
      "epoch": 0.600079904115062,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0947,
      "step": 1502
    },
    {
      "epoch": 0.6004794246903715,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0927,
      "step": 1503
    },
    {
      "epoch": 0.6008789452656812,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0922,
      "step": 1504
    },
    {
      "epoch": 0.6012784658409908,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0924,
      "step": 1505
    },
    {
      "epoch": 0.6016779864163004,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005,
      "loss": 1.1013,
      "step": 1506
    },
    {
      "epoch": 0.6020775069916101,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005,
      "loss": 1.0866,
      "step": 1507
    },
    {
      "epoch": 0.6024770275669197,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.1005,
      "step": 1508
    },
    {
      "epoch": 0.6028765481422294,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0005,
      "loss": 1.0924,
      "step": 1509
    },
    {
      "epoch": 0.6032760687175389,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0005,
      "loss": 1.1078,
      "step": 1510
    },
    {
      "epoch": 0.6036755892928486,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0806,
      "step": 1511
    },
    {
      "epoch": 0.6040751098681582,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005,
      "loss": 1.0932,
      "step": 1512
    },
    {
      "epoch": 0.6044746304434678,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0005,
      "loss": 1.0909,
      "step": 1513
    },
    {
      "epoch": 0.6048741510187775,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.086,
      "step": 1514
    },
    {
      "epoch": 0.6052736715940871,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0005,
      "loss": 1.0995,
      "step": 1515
    },
    {
      "epoch": 0.6056731921693967,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0905,
      "step": 1516
    },
    {
      "epoch": 0.6060727127447063,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0953,
      "step": 1517
    },
    {
      "epoch": 0.606472233320016,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.0861,
      "step": 1518
    },
    {
      "epoch": 0.6068717538953257,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0991,
      "step": 1519
    },
    {
      "epoch": 0.6072712744706352,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0966,
      "step": 1520
    },
    {
      "epoch": 0.6076707950459449,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0896,
      "step": 1521
    },
    {
      "epoch": 0.6080703156212545,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.0789,
      "step": 1522
    },
    {
      "epoch": 0.6084698361965641,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0953,
      "step": 1523
    },
    {
      "epoch": 0.6088693567718737,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0005,
      "loss": 1.1046,
      "step": 1524
    },
    {
      "epoch": 0.6092688773471834,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0901,
      "step": 1525
    },
    {
      "epoch": 0.609668397922493,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.1002,
      "step": 1526
    },
    {
      "epoch": 0.6100679184978026,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0978,
      "step": 1527
    },
    {
      "epoch": 0.6104674390731123,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.1081,
      "step": 1528
    },
    {
      "epoch": 0.6108669596484219,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.1032,
      "step": 1529
    },
    {
      "epoch": 0.6112664802237315,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0005,
      "loss": 1.0934,
      "step": 1530
    },
    {
      "epoch": 0.6116660007990411,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0982,
      "step": 1531
    },
    {
      "epoch": 0.6120655213743508,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0911,
      "step": 1532
    },
    {
      "epoch": 0.6124650419496604,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.094,
      "step": 1533
    },
    {
      "epoch": 0.61286456252497,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0869,
      "step": 1534
    },
    {
      "epoch": 0.6132640831002797,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0956,
      "step": 1535
    },
    {
      "epoch": 0.6136636036755893,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0867,
      "step": 1536
    },
    {
      "epoch": 0.6140631242508989,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.101,
      "step": 1537
    },
    {
      "epoch": 0.6144626448262086,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0904,
      "step": 1538
    },
    {
      "epoch": 0.6148621654015182,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0953,
      "step": 1539
    },
    {
      "epoch": 0.6152616859768278,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0948,
      "step": 1540
    },
    {
      "epoch": 0.6156612065521374,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.1022,
      "step": 1541
    },
    {
      "epoch": 0.6160607271274471,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0955,
      "step": 1542
    },
    {
      "epoch": 0.6164602477027566,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.1014,
      "step": 1543
    },
    {
      "epoch": 0.6168597682780663,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0896,
      "step": 1544
    },
    {
      "epoch": 0.617259288853376,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.097,
      "step": 1545
    },
    {
      "epoch": 0.6176588094286856,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0932,
      "step": 1546
    },
    {
      "epoch": 0.6180583300039952,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0876,
      "step": 1547
    },
    {
      "epoch": 0.6184578505793048,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0966,
      "step": 1548
    },
    {
      "epoch": 0.6188573711546145,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.1061,
      "step": 1549
    },
    {
      "epoch": 0.619256891729924,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0967,
      "step": 1550
    },
    {
      "epoch": 0.6196564123052337,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0879,
      "step": 1551
    },
    {
      "epoch": 0.6200559328805434,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0005,
      "loss": 1.1004,
      "step": 1552
    },
    {
      "epoch": 0.620455453455853,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0982,
      "step": 1553
    },
    {
      "epoch": 0.6208549740311626,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0911,
      "step": 1554
    },
    {
      "epoch": 0.6212544946064722,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0915,
      "step": 1555
    },
    {
      "epoch": 0.6216540151817819,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.098,
      "step": 1556
    },
    {
      "epoch": 0.6220535357570914,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0954,
      "step": 1557
    },
    {
      "epoch": 0.6224530563324011,
      "grad_norm": 0.171875,
      "learning_rate": 0.0005,
      "loss": 1.0986,
      "step": 1558
    },
    {
      "epoch": 0.6228525769077108,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0005,
      "loss": 1.1049,
      "step": 1559
    },
    {
      "epoch": 0.6232520974830204,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0908,
      "step": 1560
    },
    {
      "epoch": 0.62365161805833,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0907,
      "step": 1561
    },
    {
      "epoch": 0.6240511386336396,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0904,
      "step": 1562
    },
    {
      "epoch": 0.6244506592089493,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.1028,
      "step": 1563
    },
    {
      "epoch": 0.6248501797842589,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0901,
      "step": 1564
    },
    {
      "epoch": 0.6252497003595685,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0986,
      "step": 1565
    },
    {
      "epoch": 0.6256492209348782,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.1026,
      "step": 1566
    },
    {
      "epoch": 0.6260487415101877,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0999,
      "step": 1567
    },
    {
      "epoch": 0.6264482620854974,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.1026,
      "step": 1568
    },
    {
      "epoch": 0.626847782660807,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.088,
      "step": 1569
    },
    {
      "epoch": 0.6272473032361167,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.1023,
      "step": 1570
    },
    {
      "epoch": 0.6276468238114263,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005,
      "loss": 1.0978,
      "step": 1571
    },
    {
      "epoch": 0.6280463443867359,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0939,
      "step": 1572
    },
    {
      "epoch": 0.6284458649620456,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0953,
      "step": 1573
    },
    {
      "epoch": 0.6288453855373551,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0885,
      "step": 1574
    },
    {
      "epoch": 0.6292449061126648,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.1003,
      "step": 1575
    },
    {
      "epoch": 0.6296444266879744,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.1037,
      "step": 1576
    },
    {
      "epoch": 0.630043947263284,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0883,
      "step": 1577
    },
    {
      "epoch": 0.6304434678385937,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0907,
      "step": 1578
    },
    {
      "epoch": 0.6308429884139033,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.083,
      "step": 1579
    },
    {
      "epoch": 0.631242508989213,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.091,
      "step": 1580
    },
    {
      "epoch": 0.6316420295645225,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0842,
      "step": 1581
    },
    {
      "epoch": 0.6320415501398322,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0939,
      "step": 1582
    },
    {
      "epoch": 0.6324410707151419,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0862,
      "step": 1583
    },
    {
      "epoch": 0.6328405912904514,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0942,
      "step": 1584
    },
    {
      "epoch": 0.6332401118657611,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0961,
      "step": 1585
    },
    {
      "epoch": 0.6336396324410707,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0912,
      "step": 1586
    },
    {
      "epoch": 0.6340391530163804,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0802,
      "step": 1587
    },
    {
      "epoch": 0.6344386735916899,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0981,
      "step": 1588
    },
    {
      "epoch": 0.6348381941669996,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0895,
      "step": 1589
    },
    {
      "epoch": 0.6352377147423093,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0852,
      "step": 1590
    },
    {
      "epoch": 0.6356372353176188,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0893,
      "step": 1591
    },
    {
      "epoch": 0.6360367558929285,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0983,
      "step": 1592
    },
    {
      "epoch": 0.6364362764682381,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0888,
      "step": 1593
    },
    {
      "epoch": 0.6368357970435478,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0805,
      "step": 1594
    },
    {
      "epoch": 0.6372353176188573,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0914,
      "step": 1595
    },
    {
      "epoch": 0.637634838194167,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0781,
      "step": 1596
    },
    {
      "epoch": 0.6380343587694767,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0965,
      "step": 1597
    },
    {
      "epoch": 0.6384338793447862,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0882,
      "step": 1598
    },
    {
      "epoch": 0.6388333999200959,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0847,
      "step": 1599
    },
    {
      "epoch": 0.6392329204954055,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0959,
      "step": 1600
    },
    {
      "epoch": 0.6396324410707152,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.1055,
      "step": 1601
    },
    {
      "epoch": 0.6400319616460247,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0854,
      "step": 1602
    },
    {
      "epoch": 0.6404314822213344,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0912,
      "step": 1603
    },
    {
      "epoch": 0.6408310027966441,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0906,
      "step": 1604
    },
    {
      "epoch": 0.6412305233719536,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0813,
      "step": 1605
    },
    {
      "epoch": 0.6416300439472633,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.085,
      "step": 1606
    },
    {
      "epoch": 0.6420295645225729,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.097,
      "step": 1607
    },
    {
      "epoch": 0.6424290850978825,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0947,
      "step": 1608
    },
    {
      "epoch": 0.6428286056731922,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0854,
      "step": 1609
    },
    {
      "epoch": 0.6432281262485018,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0887,
      "step": 1610
    },
    {
      "epoch": 0.6436276468238115,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0005,
      "loss": 1.0946,
      "step": 1611
    },
    {
      "epoch": 0.644027167399121,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0005,
      "loss": 1.0864,
      "step": 1612
    },
    {
      "epoch": 0.6444266879744307,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0992,
      "step": 1613
    },
    {
      "epoch": 0.6448262085497403,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0005,
      "loss": 1.0935,
      "step": 1614
    },
    {
      "epoch": 0.6452257291250499,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0005,
      "loss": 1.0971,
      "step": 1615
    },
    {
      "epoch": 0.6456252497003596,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0844,
      "step": 1616
    },
    {
      "epoch": 0.6460247702756692,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005,
      "loss": 1.0843,
      "step": 1617
    },
    {
      "epoch": 0.6464242908509789,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0005,
      "loss": 1.0931,
      "step": 1618
    },
    {
      "epoch": 0.6468238114262884,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0005,
      "loss": 1.0966,
      "step": 1619
    },
    {
      "epoch": 0.6472233320015981,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005,
      "loss": 1.1025,
      "step": 1620
    },
    {
      "epoch": 0.6476228525769077,
      "grad_norm": 0.875,
      "learning_rate": 0.0005,
      "loss": 1.107,
      "step": 1621
    },
    {
      "epoch": 0.6480223731522173,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0005,
      "loss": 1.0943,
      "step": 1622
    },
    {
      "epoch": 0.648421893727527,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005,
      "loss": 1.0986,
      "step": 1623
    },
    {
      "epoch": 0.6488214143028366,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0005,
      "loss": 1.1045,
      "step": 1624
    },
    {
      "epoch": 0.6492209348781462,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0005,
      "loss": 1.0917,
      "step": 1625
    },
    {
      "epoch": 0.6496204554534558,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005,
      "loss": 1.0948,
      "step": 1626
    },
    {
      "epoch": 0.6500199760287655,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005,
      "loss": 1.0929,
      "step": 1627
    },
    {
      "epoch": 0.6504194966040752,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0005,
      "loss": 1.1023,
      "step": 1628
    },
    {
      "epoch": 0.6508190171793847,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0932,
      "step": 1629
    },
    {
      "epoch": 0.6512185377546944,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005,
      "loss": 1.0909,
      "step": 1630
    },
    {
      "epoch": 0.651618058330004,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0943,
      "step": 1631
    },
    {
      "epoch": 0.6520175789053136,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0005,
      "loss": 1.0926,
      "step": 1632
    },
    {
      "epoch": 0.6524170994806232,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0005,
      "loss": 1.0971,
      "step": 1633
    },
    {
      "epoch": 0.6528166200559329,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0893,
      "step": 1634
    },
    {
      "epoch": 0.6532161406312426,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.09,
      "step": 1635
    },
    {
      "epoch": 0.6536156612065521,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0816,
      "step": 1636
    },
    {
      "epoch": 0.6540151817818618,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0932,
      "step": 1637
    },
    {
      "epoch": 0.6544147023571714,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0918,
      "step": 1638
    },
    {
      "epoch": 0.654814222932481,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0916,
      "step": 1639
    },
    {
      "epoch": 0.6552137435077906,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.097,
      "step": 1640
    },
    {
      "epoch": 0.6556132640831003,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0899,
      "step": 1641
    },
    {
      "epoch": 0.65601278465841,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0968,
      "step": 1642
    },
    {
      "epoch": 0.6564123052337195,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0857,
      "step": 1643
    },
    {
      "epoch": 0.6568118258090292,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.086,
      "step": 1644
    },
    {
      "epoch": 0.6572113463843388,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0797,
      "step": 1645
    },
    {
      "epoch": 0.6576108669596484,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0993,
      "step": 1646
    },
    {
      "epoch": 0.658010387534958,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0876,
      "step": 1647
    },
    {
      "epoch": 0.6584099081102677,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0903,
      "step": 1648
    },
    {
      "epoch": 0.6588094286855773,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0934,
      "step": 1649
    },
    {
      "epoch": 0.6592089492608869,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0832,
      "step": 1650
    },
    {
      "epoch": 0.6596084698361966,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0845,
      "step": 1651
    },
    {
      "epoch": 0.6600079904115062,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.105,
      "step": 1652
    },
    {
      "epoch": 0.6604075109868158,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0938,
      "step": 1653
    },
    {
      "epoch": 0.6608070315621255,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0922,
      "step": 1654
    },
    {
      "epoch": 0.6612065521374351,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0904,
      "step": 1655
    },
    {
      "epoch": 0.6616060727127447,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0893,
      "step": 1656
    },
    {
      "epoch": 0.6620055932880543,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0914,
      "step": 1657
    },
    {
      "epoch": 0.662405113863364,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0895,
      "step": 1658
    },
    {
      "epoch": 0.6628046344386735,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0914,
      "step": 1659
    },
    {
      "epoch": 0.6632041550139832,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.1028,
      "step": 1660
    },
    {
      "epoch": 0.6636036755892929,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0941,
      "step": 1661
    },
    {
      "epoch": 0.6640031961646025,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0805,
      "step": 1662
    },
    {
      "epoch": 0.6644027167399121,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.096,
      "step": 1663
    },
    {
      "epoch": 0.6648022373152217,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0946,
      "step": 1664
    },
    {
      "epoch": 0.6652017578905314,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0992,
      "step": 1665
    },
    {
      "epoch": 0.6656012784658409,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.101,
      "step": 1666
    },
    {
      "epoch": 0.6660007990411506,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0934,
      "step": 1667
    },
    {
      "epoch": 0.6664003196164603,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.086,
      "step": 1668
    },
    {
      "epoch": 0.6667998401917699,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0896,
      "step": 1669
    },
    {
      "epoch": 0.6671993607670795,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0941,
      "step": 1670
    },
    {
      "epoch": 0.6675988813423891,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0865,
      "step": 1671
    },
    {
      "epoch": 0.6679984019176988,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0788,
      "step": 1672
    },
    {
      "epoch": 0.6683979224930084,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.1034,
      "step": 1673
    },
    {
      "epoch": 0.668797443068318,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0877,
      "step": 1674
    },
    {
      "epoch": 0.6691969636436277,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.1017,
      "step": 1675
    },
    {
      "epoch": 0.6695964842189372,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0869,
      "step": 1676
    },
    {
      "epoch": 0.6699960047942469,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0828,
      "step": 1677
    },
    {
      "epoch": 0.6703955253695565,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0887,
      "step": 1678
    },
    {
      "epoch": 0.6707950459448662,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0928,
      "step": 1679
    },
    {
      "epoch": 0.6711945665201758,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0943,
      "step": 1680
    },
    {
      "epoch": 0.6715940870954854,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0919,
      "step": 1681
    },
    {
      "epoch": 0.6719936076707951,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0931,
      "step": 1682
    },
    {
      "epoch": 0.6723931282461046,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0968,
      "step": 1683
    },
    {
      "epoch": 0.6727926488214143,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0833,
      "step": 1684
    },
    {
      "epoch": 0.6731921693967239,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0892,
      "step": 1685
    },
    {
      "epoch": 0.6735916899720336,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0916,
      "step": 1686
    },
    {
      "epoch": 0.6739912105473432,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0883,
      "step": 1687
    },
    {
      "epoch": 0.6743907311226528,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.095,
      "step": 1688
    },
    {
      "epoch": 0.6747902516979625,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0809,
      "step": 1689
    },
    {
      "epoch": 0.675189772273272,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0913,
      "step": 1690
    },
    {
      "epoch": 0.6755892928485817,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0855,
      "step": 1691
    },
    {
      "epoch": 0.6759888134238913,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0908,
      "step": 1692
    },
    {
      "epoch": 0.676388333999201,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0872,
      "step": 1693
    },
    {
      "epoch": 0.6767878545745106,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.0866,
      "step": 1694
    },
    {
      "epoch": 0.6771873751498202,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005,
      "loss": 1.0876,
      "step": 1695
    },
    {
      "epoch": 0.6775868957251299,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0928,
      "step": 1696
    },
    {
      "epoch": 0.6779864163004394,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005,
      "loss": 1.0782,
      "step": 1697
    },
    {
      "epoch": 0.6783859368757491,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0822,
      "step": 1698
    },
    {
      "epoch": 0.6787854574510588,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0005,
      "loss": 1.0868,
      "step": 1699
    },
    {
      "epoch": 0.6791849780263683,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0795,
      "step": 1700
    },
    {
      "epoch": 0.679584498601678,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0005,
      "loss": 1.1041,
      "step": 1701
    },
    {
      "epoch": 0.6799840191769876,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0869,
      "step": 1702
    },
    {
      "epoch": 0.6803835397522973,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0843,
      "step": 1703
    },
    {
      "epoch": 0.6807830603276068,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0989,
      "step": 1704
    },
    {
      "epoch": 0.6811825809029165,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0949,
      "step": 1705
    },
    {
      "epoch": 0.6815821014782262,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0865,
      "step": 1706
    },
    {
      "epoch": 0.6819816220535357,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0868,
      "step": 1707
    },
    {
      "epoch": 0.6823811426288454,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0856,
      "step": 1708
    },
    {
      "epoch": 0.682780663204155,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.094,
      "step": 1709
    },
    {
      "epoch": 0.6831801837794647,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.1028,
      "step": 1710
    },
    {
      "epoch": 0.6835797043547742,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0878,
      "step": 1711
    },
    {
      "epoch": 0.6839792249300839,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0833,
      "step": 1712
    },
    {
      "epoch": 0.6843787455053936,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0889,
      "step": 1713
    },
    {
      "epoch": 0.6847782660807031,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0858,
      "step": 1714
    },
    {
      "epoch": 0.6851777866560128,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0843,
      "step": 1715
    },
    {
      "epoch": 0.6855773072313224,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0927,
      "step": 1716
    },
    {
      "epoch": 0.685976827806632,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0872,
      "step": 1717
    },
    {
      "epoch": 0.6863763483819417,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0878,
      "step": 1718
    },
    {
      "epoch": 0.6867758689572513,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.1032,
      "step": 1719
    },
    {
      "epoch": 0.687175389532561,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.086,
      "step": 1720
    },
    {
      "epoch": 0.6875749101078705,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.0914,
      "step": 1721
    },
    {
      "epoch": 0.6879744306831802,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0947,
      "step": 1722
    },
    {
      "epoch": 0.6883739512584898,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0921,
      "step": 1723
    },
    {
      "epoch": 0.6887734718337994,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0864,
      "step": 1724
    },
    {
      "epoch": 0.6891729924091091,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0837,
      "step": 1725
    },
    {
      "epoch": 0.6895725129844187,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0857,
      "step": 1726
    },
    {
      "epoch": 0.6899720335597284,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0856,
      "step": 1727
    },
    {
      "epoch": 0.6903715541350379,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0814,
      "step": 1728
    },
    {
      "epoch": 0.6907710747103476,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0897,
      "step": 1729
    },
    {
      "epoch": 0.6911705952856572,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0947,
      "step": 1730
    },
    {
      "epoch": 0.6915701158609668,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0844,
      "step": 1731
    },
    {
      "epoch": 0.6919696364362765,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0978,
      "step": 1732
    },
    {
      "epoch": 0.6923691570115861,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0859,
      "step": 1733
    },
    {
      "epoch": 0.6927686775868958,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.1022,
      "step": 1734
    },
    {
      "epoch": 0.6931681981622053,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0963,
      "step": 1735
    },
    {
      "epoch": 0.693567718737515,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.097,
      "step": 1736
    },
    {
      "epoch": 0.6939672393128246,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.09,
      "step": 1737
    },
    {
      "epoch": 0.6943667598881342,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0886,
      "step": 1738
    },
    {
      "epoch": 0.6947662804634439,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0786,
      "step": 1739
    },
    {
      "epoch": 0.6951658010387535,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0782,
      "step": 1740
    },
    {
      "epoch": 0.6955653216140631,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0905,
      "step": 1741
    },
    {
      "epoch": 0.6959648421893727,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0796,
      "step": 1742
    },
    {
      "epoch": 0.6963643627646824,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0878,
      "step": 1743
    },
    {
      "epoch": 0.6967638833399921,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0872,
      "step": 1744
    },
    {
      "epoch": 0.6971634039153016,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0858,
      "step": 1745
    },
    {
      "epoch": 0.6975629244906113,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0875,
      "step": 1746
    },
    {
      "epoch": 0.6979624450659209,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0914,
      "step": 1747
    },
    {
      "epoch": 0.6983619656412305,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.091,
      "step": 1748
    },
    {
      "epoch": 0.6987614862165401,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0931,
      "step": 1749
    },
    {
      "epoch": 0.6991610067918498,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0877,
      "step": 1750
    },
    {
      "epoch": 0.6995605273671595,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0849,
      "step": 1751
    },
    {
      "epoch": 0.699960047942469,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0929,
      "step": 1752
    },
    {
      "epoch": 0.7003595685177787,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0931,
      "step": 1753
    },
    {
      "epoch": 0.7007590890930883,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0961,
      "step": 1754
    },
    {
      "epoch": 0.7011586096683979,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0867,
      "step": 1755
    },
    {
      "epoch": 0.7015581302437075,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0901,
      "step": 1756
    },
    {
      "epoch": 0.7019576508190172,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0766,
      "step": 1757
    },
    {
      "epoch": 0.7023571713943269,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0852,
      "step": 1758
    },
    {
      "epoch": 0.7027566919696364,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0873,
      "step": 1759
    },
    {
      "epoch": 0.7031562125449461,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.103,
      "step": 1760
    },
    {
      "epoch": 0.7035557331202557,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0887,
      "step": 1761
    },
    {
      "epoch": 0.7039552536955653,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0853,
      "step": 1762
    },
    {
      "epoch": 0.7043547742708749,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.1039,
      "step": 1763
    },
    {
      "epoch": 0.7047542948461846,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.1098,
      "step": 1764
    },
    {
      "epoch": 0.7051538154214942,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0897,
      "step": 1765
    },
    {
      "epoch": 0.7055533359968038,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.1023,
      "step": 1766
    },
    {
      "epoch": 0.7059528565721135,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0966,
      "step": 1767
    },
    {
      "epoch": 0.706352377147423,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0811,
      "step": 1768
    },
    {
      "epoch": 0.7067518977227327,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0986,
      "step": 1769
    },
    {
      "epoch": 0.7071514182980424,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 1770
    },
    {
      "epoch": 0.707550938873352,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.1,
      "step": 1771
    },
    {
      "epoch": 0.7079504594486616,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.095,
      "step": 1772
    },
    {
      "epoch": 0.7083499800239712,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.092,
      "step": 1773
    },
    {
      "epoch": 0.7087495005992809,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0965,
      "step": 1774
    },
    {
      "epoch": 0.7091490211745904,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0936,
      "step": 1775
    },
    {
      "epoch": 0.7095485417499001,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 1776
    },
    {
      "epoch": 0.7099480623252098,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0864,
      "step": 1777
    },
    {
      "epoch": 0.7103475829005194,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0899,
      "step": 1778
    },
    {
      "epoch": 0.710747103475829,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.088,
      "step": 1779
    },
    {
      "epoch": 0.7111466240511386,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0931,
      "step": 1780
    },
    {
      "epoch": 0.7115461446264483,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0839,
      "step": 1781
    },
    {
      "epoch": 0.7119456652017578,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0911,
      "step": 1782
    },
    {
      "epoch": 0.7123451857770675,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0869,
      "step": 1783
    },
    {
      "epoch": 0.7127447063523772,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0913,
      "step": 1784
    },
    {
      "epoch": 0.7131442269276868,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0944,
      "step": 1785
    },
    {
      "epoch": 0.7135437475029964,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0005,
      "loss": 1.0859,
      "step": 1786
    },
    {
      "epoch": 0.713943268078306,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0906,
      "step": 1787
    },
    {
      "epoch": 0.7143427886536157,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0925,
      "step": 1788
    },
    {
      "epoch": 0.7147423092289253,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.093,
      "step": 1789
    },
    {
      "epoch": 0.7151418298042349,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.1006,
      "step": 1790
    },
    {
      "epoch": 0.7155413503795446,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0951,
      "step": 1791
    },
    {
      "epoch": 0.7159408709548541,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0849,
      "step": 1792
    },
    {
      "epoch": 0.7163403915301638,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0898,
      "step": 1793
    },
    {
      "epoch": 0.7167399121054734,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0902,
      "step": 1794
    },
    {
      "epoch": 0.7171394326807831,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0855,
      "step": 1795
    },
    {
      "epoch": 0.7175389532560927,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0902,
      "step": 1796
    },
    {
      "epoch": 0.7179384738314023,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0839,
      "step": 1797
    },
    {
      "epoch": 0.718337994406712,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0937,
      "step": 1798
    },
    {
      "epoch": 0.7187375149820215,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0934,
      "step": 1799
    },
    {
      "epoch": 0.7191370355573312,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0861,
      "step": 1800
    },
    {
      "epoch": 0.7195365561326408,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0947,
      "step": 1801
    },
    {
      "epoch": 0.7199360767079505,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0806,
      "step": 1802
    },
    {
      "epoch": 0.7203355972832601,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.086,
      "step": 1803
    },
    {
      "epoch": 0.7207351178585697,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0882,
      "step": 1804
    },
    {
      "epoch": 0.7211346384338794,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0816,
      "step": 1805
    },
    {
      "epoch": 0.7215341590091889,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0939,
      "step": 1806
    },
    {
      "epoch": 0.7219336795844986,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0937,
      "step": 1807
    },
    {
      "epoch": 0.7223332001598082,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0893,
      "step": 1808
    },
    {
      "epoch": 0.7227327207351179,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0005,
      "loss": 1.0904,
      "step": 1809
    },
    {
      "epoch": 0.7231322413104275,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0972,
      "step": 1810
    },
    {
      "epoch": 0.7235317618857371,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0956,
      "step": 1811
    },
    {
      "epoch": 0.7239312824610468,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0768,
      "step": 1812
    },
    {
      "epoch": 0.7243308030363563,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0913,
      "step": 1813
    },
    {
      "epoch": 0.724730323611666,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.073,
      "step": 1814
    },
    {
      "epoch": 0.7251298441869757,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0935,
      "step": 1815
    },
    {
      "epoch": 0.7255293647622852,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0947,
      "step": 1816
    },
    {
      "epoch": 0.7259288853375949,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0919,
      "step": 1817
    },
    {
      "epoch": 0.7263284059129045,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.1062,
      "step": 1818
    },
    {
      "epoch": 0.7267279264882142,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0973,
      "step": 1819
    },
    {
      "epoch": 0.7271274470635237,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0838,
      "step": 1820
    },
    {
      "epoch": 0.7275269676388334,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0989,
      "step": 1821
    },
    {
      "epoch": 0.7279264882141431,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 1822
    },
    {
      "epoch": 0.7283260087894526,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0996,
      "step": 1823
    },
    {
      "epoch": 0.7287255293647623,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0816,
      "step": 1824
    },
    {
      "epoch": 0.7291250499400719,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0825,
      "step": 1825
    },
    {
      "epoch": 0.7295245705153816,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0779,
      "step": 1826
    },
    {
      "epoch": 0.7299240910906911,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.082,
      "step": 1827
    },
    {
      "epoch": 0.7303236116660008,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0846,
      "step": 1828
    },
    {
      "epoch": 0.7307231322413105,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0938,
      "step": 1829
    },
    {
      "epoch": 0.73112265281662,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0832,
      "step": 1830
    },
    {
      "epoch": 0.7315221733919297,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0874,
      "step": 1831
    },
    {
      "epoch": 0.7319216939672393,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.1091,
      "step": 1832
    },
    {
      "epoch": 0.732321214542549,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0997,
      "step": 1833
    },
    {
      "epoch": 0.7327207351178586,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0951,
      "step": 1834
    },
    {
      "epoch": 0.7331202556931682,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0716,
      "step": 1835
    },
    {
      "epoch": 0.7335197762684779,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.08,
      "step": 1836
    },
    {
      "epoch": 0.7339192968437874,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0888,
      "step": 1837
    },
    {
      "epoch": 0.7343188174190971,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.09,
      "step": 1838
    },
    {
      "epoch": 0.7347183379944067,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0845,
      "step": 1839
    },
    {
      "epoch": 0.7351178585697163,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0965,
      "step": 1840
    },
    {
      "epoch": 0.735517379145026,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0868,
      "step": 1841
    },
    {
      "epoch": 0.7359168997203356,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.101,
      "step": 1842
    },
    {
      "epoch": 0.7363164202956453,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0925,
      "step": 1843
    },
    {
      "epoch": 0.7367159408709548,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0892,
      "step": 1844
    },
    {
      "epoch": 0.7371154614462645,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0839,
      "step": 1845
    },
    {
      "epoch": 0.7375149820215741,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0842,
      "step": 1846
    },
    {
      "epoch": 0.7379145025968837,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0852,
      "step": 1847
    },
    {
      "epoch": 0.7383140231721934,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0873,
      "step": 1848
    },
    {
      "epoch": 0.738713543747503,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 1849
    },
    {
      "epoch": 0.7391130643228127,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0809,
      "step": 1850
    },
    {
      "epoch": 0.7395125848981222,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0968,
      "step": 1851
    },
    {
      "epoch": 0.7399121054734319,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0882,
      "step": 1852
    },
    {
      "epoch": 0.7403116260487415,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0854,
      "step": 1853
    },
    {
      "epoch": 0.7407111466240511,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.084,
      "step": 1854
    },
    {
      "epoch": 0.7411106671993608,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0856,
      "step": 1855
    },
    {
      "epoch": 0.7415101877746704,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0927,
      "step": 1856
    },
    {
      "epoch": 0.74190970834998,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0963,
      "step": 1857
    },
    {
      "epoch": 0.7423092289252896,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0929,
      "step": 1858
    },
    {
      "epoch": 0.7427087495005993,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0802,
      "step": 1859
    },
    {
      "epoch": 0.743108270075909,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.1038,
      "step": 1860
    },
    {
      "epoch": 0.7435077906512185,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0961,
      "step": 1861
    },
    {
      "epoch": 0.7439073112265282,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0875,
      "step": 1862
    },
    {
      "epoch": 0.7443068318018378,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0817,
      "step": 1863
    },
    {
      "epoch": 0.7447063523771474,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.085,
      "step": 1864
    },
    {
      "epoch": 0.745105872952457,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0754,
      "step": 1865
    },
    {
      "epoch": 0.7455053935277667,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0903,
      "step": 1866
    },
    {
      "epoch": 0.7459049141030764,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 1867
    },
    {
      "epoch": 0.7463044346783859,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0997,
      "step": 1868
    },
    {
      "epoch": 0.7467039552536956,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0868,
      "step": 1869
    },
    {
      "epoch": 0.7471034758290052,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0792,
      "step": 1870
    },
    {
      "epoch": 0.7475029964043148,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0794,
      "step": 1871
    },
    {
      "epoch": 0.7479025169796244,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0917,
      "step": 1872
    },
    {
      "epoch": 0.7483020375549341,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0927,
      "step": 1873
    },
    {
      "epoch": 0.7487015581302437,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.1003,
      "step": 1874
    },
    {
      "epoch": 0.7491010787055533,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0866,
      "step": 1875
    },
    {
      "epoch": 0.749500599280863,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0788,
      "step": 1876
    },
    {
      "epoch": 0.7499001198561726,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0861,
      "step": 1877
    },
    {
      "epoch": 0.7502996404314822,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0888,
      "step": 1878
    },
    {
      "epoch": 0.7506991610067919,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0886,
      "step": 1879
    },
    {
      "epoch": 0.7510986815821015,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0917,
      "step": 1880
    },
    {
      "epoch": 0.7514982021574111,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0819,
      "step": 1881
    },
    {
      "epoch": 0.7518977227327207,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0906,
      "step": 1882
    },
    {
      "epoch": 0.7522972433080304,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0913,
      "step": 1883
    },
    {
      "epoch": 0.75269676388334,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0928,
      "step": 1884
    },
    {
      "epoch": 0.7530962844586496,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0871,
      "step": 1885
    },
    {
      "epoch": 0.7534958050339593,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0905,
      "step": 1886
    },
    {
      "epoch": 0.7538953256092689,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0856,
      "step": 1887
    },
    {
      "epoch": 0.7542948461845785,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0005,
      "loss": 1.0862,
      "step": 1888
    },
    {
      "epoch": 0.7546943667598881,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.1018,
      "step": 1889
    },
    {
      "epoch": 0.7550938873351978,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0868,
      "step": 1890
    },
    {
      "epoch": 0.7554934079105073,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0912,
      "step": 1891
    },
    {
      "epoch": 0.755892928485817,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0915,
      "step": 1892
    },
    {
      "epoch": 0.7562924490611267,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0903,
      "step": 1893
    },
    {
      "epoch": 0.7566919696364363,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0727,
      "step": 1894
    },
    {
      "epoch": 0.7570914902117459,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.089,
      "step": 1895
    },
    {
      "epoch": 0.7574910107870555,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0875,
      "step": 1896
    },
    {
      "epoch": 0.7578905313623652,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0734,
      "step": 1897
    },
    {
      "epoch": 0.7582900519376747,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0767,
      "step": 1898
    },
    {
      "epoch": 0.7586895725129844,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.087,
      "step": 1899
    },
    {
      "epoch": 0.7590890930882941,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0863,
      "step": 1900
    },
    {
      "epoch": 0.7594886136636037,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0975,
      "step": 1901
    },
    {
      "epoch": 0.7598881342389133,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0914,
      "step": 1902
    },
    {
      "epoch": 0.7602876548142229,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0862,
      "step": 1903
    },
    {
      "epoch": 0.7606871753895326,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0906,
      "step": 1904
    },
    {
      "epoch": 0.7610866959648422,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0711,
      "step": 1905
    },
    {
      "epoch": 0.7614862165401518,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0897,
      "step": 1906
    },
    {
      "epoch": 0.7618857371154615,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.1093,
      "step": 1907
    },
    {
      "epoch": 0.762285257690771,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0735,
      "step": 1908
    },
    {
      "epoch": 0.7626847782660807,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0915,
      "step": 1909
    },
    {
      "epoch": 0.7630842988413903,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0765,
      "step": 1910
    },
    {
      "epoch": 0.7634838194167,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0739,
      "step": 1911
    },
    {
      "epoch": 0.7638833399920096,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.088,
      "step": 1912
    },
    {
      "epoch": 0.7642828605673192,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0844,
      "step": 1913
    },
    {
      "epoch": 0.7646823811426289,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0909,
      "step": 1914
    },
    {
      "epoch": 0.7650819017179384,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0955,
      "step": 1915
    },
    {
      "epoch": 0.7654814222932481,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0849,
      "step": 1916
    },
    {
      "epoch": 0.7658809428685577,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0849,
      "step": 1917
    },
    {
      "epoch": 0.7662804634438674,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0757,
      "step": 1918
    },
    {
      "epoch": 0.766679984019177,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0771,
      "step": 1919
    },
    {
      "epoch": 0.7670795045944866,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0943,
      "step": 1920
    },
    {
      "epoch": 0.7674790251697963,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0941,
      "step": 1921
    },
    {
      "epoch": 0.7678785457451058,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0837,
      "step": 1922
    },
    {
      "epoch": 0.7682780663204155,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0941,
      "step": 1923
    },
    {
      "epoch": 0.7686775868957252,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0855,
      "step": 1924
    },
    {
      "epoch": 0.7690771074710347,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0827,
      "step": 1925
    },
    {
      "epoch": 0.7694766280463444,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.08,
      "step": 1926
    },
    {
      "epoch": 0.769876148621654,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0005,
      "loss": 1.0795,
      "step": 1927
    },
    {
      "epoch": 0.7702756691969637,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0005,
      "loss": 1.1004,
      "step": 1928
    },
    {
      "epoch": 0.7706751897722732,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.087,
      "step": 1929
    },
    {
      "epoch": 0.7710747103475829,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0005,
      "loss": 1.0984,
      "step": 1930
    },
    {
      "epoch": 0.7714742309228926,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0787,
      "step": 1931
    },
    {
      "epoch": 0.7718737514982021,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0867,
      "step": 1932
    },
    {
      "epoch": 0.7722732720735118,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0005,
      "loss": 1.0887,
      "step": 1933
    },
    {
      "epoch": 0.7726727926488214,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0965,
      "step": 1934
    },
    {
      "epoch": 0.7730723132241311,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.0912,
      "step": 1935
    },
    {
      "epoch": 0.7734718337994406,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.0977,
      "step": 1936
    },
    {
      "epoch": 0.7738713543747503,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0968,
      "step": 1937
    },
    {
      "epoch": 0.77427087495006,
      "grad_norm": 0.296875,
      "learning_rate": 0.0005,
      "loss": 1.1095,
      "step": 1938
    },
    {
      "epoch": 0.7746703955253695,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0005,
      "loss": 1.0862,
      "step": 1939
    },
    {
      "epoch": 0.7750699161006792,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 1940
    },
    {
      "epoch": 0.7754694366759888,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005,
      "loss": 1.0992,
      "step": 1941
    },
    {
      "epoch": 0.7758689572512985,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 1942
    },
    {
      "epoch": 0.776268477826608,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0777,
      "step": 1943
    },
    {
      "epoch": 0.7766679984019177,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0849,
      "step": 1944
    },
    {
      "epoch": 0.7770675189772274,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0837,
      "step": 1945
    },
    {
      "epoch": 0.7774670395525369,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0005,
      "loss": 1.088,
      "step": 1946
    },
    {
      "epoch": 0.7778665601278466,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0932,
      "step": 1947
    },
    {
      "epoch": 0.7782660807031562,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0005,
      "loss": 1.0831,
      "step": 1948
    },
    {
      "epoch": 0.7786656012784658,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0005,
      "loss": 1.0836,
      "step": 1949
    },
    {
      "epoch": 0.7790651218537755,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0799,
      "step": 1950
    },
    {
      "epoch": 0.7794646424290851,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0005,
      "loss": 1.0835,
      "step": 1951
    },
    {
      "epoch": 0.7798641630043948,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 1952
    },
    {
      "epoch": 0.7802636835797043,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.0822,
      "step": 1953
    },
    {
      "epoch": 0.780663204155014,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0005,
      "loss": 1.088,
      "step": 1954
    },
    {
      "epoch": 0.7810627247303236,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0867,
      "step": 1955
    },
    {
      "epoch": 0.7814622453056332,
      "grad_norm": 0.234375,
      "learning_rate": 0.0005,
      "loss": 1.0961,
      "step": 1956
    },
    {
      "epoch": 0.7818617658809429,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.08,
      "step": 1957
    },
    {
      "epoch": 0.7822612864562525,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0851,
      "step": 1958
    },
    {
      "epoch": 0.7826608070315622,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.0864,
      "step": 1959
    },
    {
      "epoch": 0.7830603276068717,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0911,
      "step": 1960
    },
    {
      "epoch": 0.7834598481821814,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0856,
      "step": 1961
    },
    {
      "epoch": 0.783859368757491,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0815,
      "step": 1962
    },
    {
      "epoch": 0.7842588893328006,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0996,
      "step": 1963
    },
    {
      "epoch": 0.7846584099081103,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0886,
      "step": 1964
    },
    {
      "epoch": 0.7850579304834199,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.085,
      "step": 1965
    },
    {
      "epoch": 0.7854574510587295,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0944,
      "step": 1966
    },
    {
      "epoch": 0.7858569716340391,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0765,
      "step": 1967
    },
    {
      "epoch": 0.7862564922093488,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0745,
      "step": 1968
    },
    {
      "epoch": 0.7866560127846585,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.0739,
      "step": 1969
    },
    {
      "epoch": 0.787055533359968,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0776,
      "step": 1970
    },
    {
      "epoch": 0.7874550539352777,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0923,
      "step": 1971
    },
    {
      "epoch": 0.7878545745105873,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.085,
      "step": 1972
    },
    {
      "epoch": 0.7882540950858969,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.0941,
      "step": 1973
    },
    {
      "epoch": 0.7886536156612065,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0807,
      "step": 1974
    },
    {
      "epoch": 0.7890531362365162,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.074,
      "step": 1975
    },
    {
      "epoch": 0.7894526568118259,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0816,
      "step": 1976
    },
    {
      "epoch": 0.7898521773871354,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0005,
      "loss": 1.0874,
      "step": 1977
    },
    {
      "epoch": 0.7902516979624451,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 1978
    },
    {
      "epoch": 0.7906512185377547,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.078,
      "step": 1979
    },
    {
      "epoch": 0.7910507391130643,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0005,
      "loss": 1.0898,
      "step": 1980
    },
    {
      "epoch": 0.7914502596883739,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0005,
      "loss": 1.0811,
      "step": 1981
    },
    {
      "epoch": 0.7918497802636836,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0005,
      "loss": 1.0792,
      "step": 1982
    },
    {
      "epoch": 0.7922493008389933,
      "grad_norm": 0.5,
      "learning_rate": 0.0005,
      "loss": 1.0871,
      "step": 1983
    },
    {
      "epoch": 0.7926488214143028,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0005,
      "loss": 1.0914,
      "step": 1984
    },
    {
      "epoch": 0.7930483419896125,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0005,
      "loss": 1.0892,
      "step": 1985
    },
    {
      "epoch": 0.7934478625649221,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0899,
      "step": 1986
    },
    {
      "epoch": 0.7938473831402317,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0005,
      "loss": 1.095,
      "step": 1987
    },
    {
      "epoch": 0.7942469037155413,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0005,
      "loss": 1.1058,
      "step": 1988
    },
    {
      "epoch": 0.794646424290851,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0769,
      "step": 1989
    },
    {
      "epoch": 0.7950459448661606,
      "grad_norm": 0.5625,
      "learning_rate": 0.0005,
      "loss": 1.0984,
      "step": 1990
    },
    {
      "epoch": 0.7954454654414702,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0005,
      "loss": 1.0907,
      "step": 1991
    },
    {
      "epoch": 0.7958449860167799,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0005,
      "loss": 1.0977,
      "step": 1992
    },
    {
      "epoch": 0.7962445065920895,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0005,
      "loss": 1.0874,
      "step": 1993
    },
    {
      "epoch": 0.7966440271673991,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005,
      "loss": 1.0917,
      "step": 1994
    },
    {
      "epoch": 0.7970435477427088,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005,
      "loss": 1.0951,
      "step": 1995
    },
    {
      "epoch": 0.7974430683180184,
      "grad_norm": 0.578125,
      "learning_rate": 0.0005,
      "loss": 1.0841,
      "step": 1996
    },
    {
      "epoch": 0.797842588893328,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0005,
      "loss": 1.0902,
      "step": 1997
    },
    {
      "epoch": 0.7982421094686376,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0005,
      "loss": 1.0896,
      "step": 1998
    },
    {
      "epoch": 0.7986416300439473,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.095,
      "step": 1999
    },
    {
      "epoch": 0.7990411506192568,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.0851,
      "step": 2000
    },
    {
      "epoch": 0.7994406711945665,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0868,
      "step": 2001
    },
    {
      "epoch": 0.7998401917698762,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0005,
      "loss": 1.0939,
      "step": 2002
    },
    {
      "epoch": 0.8002397123451858,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0795,
      "step": 2003
    },
    {
      "epoch": 0.8006392329204954,
      "grad_norm": 0.28125,
      "learning_rate": 0.0005,
      "loss": 1.0838,
      "step": 2004
    },
    {
      "epoch": 0.801038753495805,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.1013,
      "step": 2005
    },
    {
      "epoch": 0.8014382740711147,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0911,
      "step": 2006
    },
    {
      "epoch": 0.8018377946464242,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0904,
      "step": 2007
    },
    {
      "epoch": 0.8022373152217339,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 2008
    },
    {
      "epoch": 0.8026368357970436,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0838,
      "step": 2009
    },
    {
      "epoch": 0.8030363563723532,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0952,
      "step": 2010
    },
    {
      "epoch": 0.8034358769476628,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0879,
      "step": 2011
    },
    {
      "epoch": 0.8038353975229724,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0883,
      "step": 2012
    },
    {
      "epoch": 0.8042349180982821,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0831,
      "step": 2013
    },
    {
      "epoch": 0.8046344386735917,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 2014
    },
    {
      "epoch": 0.8050339592489013,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0976,
      "step": 2015
    },
    {
      "epoch": 0.805433479824211,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0836,
      "step": 2016
    },
    {
      "epoch": 0.8058330003995205,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0794,
      "step": 2017
    },
    {
      "epoch": 0.8062325209748302,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0874,
      "step": 2018
    },
    {
      "epoch": 0.8066320415501398,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0845,
      "step": 2019
    },
    {
      "epoch": 0.8070315621254495,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0849,
      "step": 2020
    },
    {
      "epoch": 0.8074310827007591,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0862,
      "step": 2021
    },
    {
      "epoch": 0.8078306032760687,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0815,
      "step": 2022
    },
    {
      "epoch": 0.8082301238513784,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0888,
      "step": 2023
    },
    {
      "epoch": 0.8086296444266879,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.081,
      "step": 2024
    },
    {
      "epoch": 0.8090291650019976,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0764,
      "step": 2025
    },
    {
      "epoch": 0.8094286855773072,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0961,
      "step": 2026
    },
    {
      "epoch": 0.8098282061526169,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0746,
      "step": 2027
    },
    {
      "epoch": 0.8102277267279265,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0825,
      "step": 2028
    },
    {
      "epoch": 0.8106272473032361,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0887,
      "step": 2029
    },
    {
      "epoch": 0.8110267678785458,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 2030
    },
    {
      "epoch": 0.8114262884538553,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0876,
      "step": 2031
    },
    {
      "epoch": 0.811825809029165,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0951,
      "step": 2032
    },
    {
      "epoch": 0.8122253296044746,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0774,
      "step": 2033
    },
    {
      "epoch": 0.8126248501797843,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0878,
      "step": 2034
    },
    {
      "epoch": 0.8130243707550939,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 2035
    },
    {
      "epoch": 0.8134238913304035,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.087,
      "step": 2036
    },
    {
      "epoch": 0.8138234119057132,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0884,
      "step": 2037
    },
    {
      "epoch": 0.8142229324810227,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0835,
      "step": 2038
    },
    {
      "epoch": 0.8146224530563324,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0769,
      "step": 2039
    },
    {
      "epoch": 0.8150219736316421,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0869,
      "step": 2040
    },
    {
      "epoch": 0.8154214942069516,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0846,
      "step": 2041
    },
    {
      "epoch": 0.8158210147822613,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0856,
      "step": 2042
    },
    {
      "epoch": 0.8162205353575709,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0837,
      "step": 2043
    },
    {
      "epoch": 0.8166200559328806,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0767,
      "step": 2044
    },
    {
      "epoch": 0.8170195765081901,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0852,
      "step": 2045
    },
    {
      "epoch": 0.8174190970834998,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0845,
      "step": 2046
    },
    {
      "epoch": 0.8178186176588095,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0823,
      "step": 2047
    },
    {
      "epoch": 0.818218138234119,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0793,
      "step": 2048
    },
    {
      "epoch": 0.8186176588094287,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.087,
      "step": 2049
    },
    {
      "epoch": 0.8190171793847383,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0968,
      "step": 2050
    },
    {
      "epoch": 0.819416699960048,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0741,
      "step": 2051
    },
    {
      "epoch": 0.8198162205353575,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0837,
      "step": 2052
    },
    {
      "epoch": 0.8202157411106672,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0704,
      "step": 2053
    },
    {
      "epoch": 0.8206152616859769,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0886,
      "step": 2054
    },
    {
      "epoch": 0.8210147822612864,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0832,
      "step": 2055
    },
    {
      "epoch": 0.8214143028365961,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0897,
      "step": 2056
    },
    {
      "epoch": 0.8218138234119057,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.087,
      "step": 2057
    },
    {
      "epoch": 0.8222133439872154,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0808,
      "step": 2058
    },
    {
      "epoch": 0.8226128645625249,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0795,
      "step": 2059
    },
    {
      "epoch": 0.8230123851378346,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 2060
    },
    {
      "epoch": 0.8234119057131443,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.084,
      "step": 2061
    },
    {
      "epoch": 0.8238114262884538,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0767,
      "step": 2062
    },
    {
      "epoch": 0.8242109468637635,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0872,
      "step": 2063
    },
    {
      "epoch": 0.8246104674390731,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0919,
      "step": 2064
    },
    {
      "epoch": 0.8250099880143827,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0769,
      "step": 2065
    },
    {
      "epoch": 0.8254095085896924,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0915,
      "step": 2066
    },
    {
      "epoch": 0.825809029165002,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0881,
      "step": 2067
    },
    {
      "epoch": 0.8262085497403117,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 2068
    },
    {
      "epoch": 0.8266080703156212,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.085,
      "step": 2069
    },
    {
      "epoch": 0.8270075908909309,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0905,
      "step": 2070
    },
    {
      "epoch": 0.8274071114662405,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0784,
      "step": 2071
    },
    {
      "epoch": 0.8278066320415501,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0834,
      "step": 2072
    },
    {
      "epoch": 0.8282061526168598,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0925,
      "step": 2073
    },
    {
      "epoch": 0.8286056731921694,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0958,
      "step": 2074
    },
    {
      "epoch": 0.829005193767479,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.09,
      "step": 2075
    },
    {
      "epoch": 0.8294047143427886,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0837,
      "step": 2076
    },
    {
      "epoch": 0.8298042349180983,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0826,
      "step": 2077
    },
    {
      "epoch": 0.8302037554934079,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0937,
      "step": 2078
    },
    {
      "epoch": 0.8306032760687175,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0793,
      "step": 2079
    },
    {
      "epoch": 0.8310027966440272,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0863,
      "step": 2080
    },
    {
      "epoch": 0.8314023172193368,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0884,
      "step": 2081
    },
    {
      "epoch": 0.8318018377946464,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0738,
      "step": 2082
    },
    {
      "epoch": 0.832201358369956,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0779,
      "step": 2083
    },
    {
      "epoch": 0.8326008789452657,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.084,
      "step": 2084
    },
    {
      "epoch": 0.8330003995205754,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0868,
      "step": 2085
    },
    {
      "epoch": 0.8333999200958849,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0744,
      "step": 2086
    },
    {
      "epoch": 0.8337994406711946,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0815,
      "step": 2087
    },
    {
      "epoch": 0.8341989612465042,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 2088
    },
    {
      "epoch": 0.8345984818218138,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0907,
      "step": 2089
    },
    {
      "epoch": 0.8349980023971234,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0955,
      "step": 2090
    },
    {
      "epoch": 0.8353975229724331,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0866,
      "step": 2091
    },
    {
      "epoch": 0.8357970435477428,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0832,
      "step": 2092
    },
    {
      "epoch": 0.8361965641230523,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0925,
      "step": 2093
    },
    {
      "epoch": 0.836596084698362,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0895,
      "step": 2094
    },
    {
      "epoch": 0.8369956052736716,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0817,
      "step": 2095
    },
    {
      "epoch": 0.8373951258489812,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0738,
      "step": 2096
    },
    {
      "epoch": 0.8377946464242908,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0974,
      "step": 2097
    },
    {
      "epoch": 0.8381941669996005,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0869,
      "step": 2098
    },
    {
      "epoch": 0.8385936875749102,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0869,
      "step": 2099
    },
    {
      "epoch": 0.8389932081502197,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0887,
      "step": 2100
    },
    {
      "epoch": 0.8393927287255294,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0882,
      "step": 2101
    },
    {
      "epoch": 0.839792249300839,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0836,
      "step": 2102
    },
    {
      "epoch": 0.8401917698761486,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0925,
      "step": 2103
    },
    {
      "epoch": 0.8405912904514582,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 2104
    },
    {
      "epoch": 0.8409908110267679,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0924,
      "step": 2105
    },
    {
      "epoch": 0.8413903316020775,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.079,
      "step": 2106
    },
    {
      "epoch": 0.8417898521773871,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.092,
      "step": 2107
    },
    {
      "epoch": 0.8421893727526968,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0793,
      "step": 2108
    },
    {
      "epoch": 0.8425888933280063,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0852,
      "step": 2109
    },
    {
      "epoch": 0.842988413903316,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.074,
      "step": 2110
    },
    {
      "epoch": 0.8433879344786257,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.084,
      "step": 2111
    },
    {
      "epoch": 0.8437874550539353,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0887,
      "step": 2112
    },
    {
      "epoch": 0.8441869756292449,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0807,
      "step": 2113
    },
    {
      "epoch": 0.8445864962045545,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0824,
      "step": 2114
    },
    {
      "epoch": 0.8449860167798642,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0788,
      "step": 2115
    },
    {
      "epoch": 0.8453855373551737,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0877,
      "step": 2116
    },
    {
      "epoch": 0.8457850579304834,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0891,
      "step": 2117
    },
    {
      "epoch": 0.8461845785057931,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0833,
      "step": 2118
    },
    {
      "epoch": 0.8465840990811027,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0821,
      "step": 2119
    },
    {
      "epoch": 0.8469836196564123,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0789,
      "step": 2120
    },
    {
      "epoch": 0.8473831402317219,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0919,
      "step": 2121
    },
    {
      "epoch": 0.8477826608070316,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0806,
      "step": 2122
    },
    {
      "epoch": 0.8481821813823411,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0871,
      "step": 2123
    },
    {
      "epoch": 0.8485817019576508,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0882,
      "step": 2124
    },
    {
      "epoch": 0.8489812225329605,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 2125
    },
    {
      "epoch": 0.84938074310827,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 2126
    },
    {
      "epoch": 0.8497802636835797,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0862,
      "step": 2127
    },
    {
      "epoch": 0.8501797842588893,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0826,
      "step": 2128
    },
    {
      "epoch": 0.850579304834199,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0861,
      "step": 2129
    },
    {
      "epoch": 0.8509788254095086,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0835,
      "step": 2130
    },
    {
      "epoch": 0.8513783459848182,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0821,
      "step": 2131
    },
    {
      "epoch": 0.8517778665601279,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0863,
      "step": 2132
    },
    {
      "epoch": 0.8521773871354374,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0832,
      "step": 2133
    },
    {
      "epoch": 0.8525769077107471,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0875,
      "step": 2134
    },
    {
      "epoch": 0.8529764282860567,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0005,
      "loss": 1.088,
      "step": 2135
    },
    {
      "epoch": 0.8533759488613664,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0005,
      "loss": 1.0807,
      "step": 2136
    },
    {
      "epoch": 0.853775469436676,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0765,
      "step": 2137
    },
    {
      "epoch": 0.8541749900119856,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005,
      "loss": 1.083,
      "step": 2138
    },
    {
      "epoch": 0.8545745105872953,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0005,
      "loss": 1.0794,
      "step": 2139
    },
    {
      "epoch": 0.8549740311626048,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0005,
      "loss": 1.0901,
      "step": 2140
    },
    {
      "epoch": 0.8553735517379145,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0005,
      "loss": 1.093,
      "step": 2141
    },
    {
      "epoch": 0.8557730723132241,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0831,
      "step": 2142
    },
    {
      "epoch": 0.8561725928885338,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0005,
      "loss": 1.1007,
      "step": 2143
    },
    {
      "epoch": 0.8565721134638434,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0848,
      "step": 2144
    },
    {
      "epoch": 0.856971634039153,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0005,
      "loss": 1.086,
      "step": 2145
    },
    {
      "epoch": 0.8573711546144627,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0921,
      "step": 2146
    },
    {
      "epoch": 0.8577706751897722,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0005,
      "loss": 1.0766,
      "step": 2147
    },
    {
      "epoch": 0.8581701957650819,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0005,
      "loss": 1.0977,
      "step": 2148
    },
    {
      "epoch": 0.8585697163403915,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0913,
      "step": 2149
    },
    {
      "epoch": 0.8589692369157012,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.0813,
      "step": 2150
    },
    {
      "epoch": 0.8593687574910108,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0754,
      "step": 2151
    },
    {
      "epoch": 0.8597682780663204,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0005,
      "loss": 1.0767,
      "step": 2152
    },
    {
      "epoch": 0.8601677986416301,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.1041,
      "step": 2153
    },
    {
      "epoch": 0.8605673192169396,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0961,
      "step": 2154
    },
    {
      "epoch": 0.8609668397922493,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0832,
      "step": 2155
    },
    {
      "epoch": 0.861366360367559,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0791,
      "step": 2156
    },
    {
      "epoch": 0.8617658809428685,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0842,
      "step": 2157
    },
    {
      "epoch": 0.8621654015181782,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 2158
    },
    {
      "epoch": 0.8625649220934878,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0772,
      "step": 2159
    },
    {
      "epoch": 0.8629644426687975,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0696,
      "step": 2160
    },
    {
      "epoch": 0.863363963244107,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0765,
      "step": 2161
    },
    {
      "epoch": 0.8637634838194167,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0795,
      "step": 2162
    },
    {
      "epoch": 0.8641630043947264,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.083,
      "step": 2163
    },
    {
      "epoch": 0.8645625249700359,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 2164
    },
    {
      "epoch": 0.8649620455453456,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0814,
      "step": 2165
    },
    {
      "epoch": 0.8653615661206552,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0835,
      "step": 2166
    },
    {
      "epoch": 0.8657610866959649,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0808,
      "step": 2167
    },
    {
      "epoch": 0.8661606072712744,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.075,
      "step": 2168
    },
    {
      "epoch": 0.8665601278465841,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0796,
      "step": 2169
    },
    {
      "epoch": 0.8669596484218938,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 2170
    },
    {
      "epoch": 0.8673591689972033,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0711,
      "step": 2171
    },
    {
      "epoch": 0.867758689572513,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0693,
      "step": 2172
    },
    {
      "epoch": 0.8681582101478226,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0822,
      "step": 2173
    },
    {
      "epoch": 0.8685577307231322,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0897,
      "step": 2174
    },
    {
      "epoch": 0.8689572512984419,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0863,
      "step": 2175
    },
    {
      "epoch": 0.8693567718737515,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 2176
    },
    {
      "epoch": 0.8697562924490612,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.078,
      "step": 2177
    },
    {
      "epoch": 0.8701558130243707,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 2178
    },
    {
      "epoch": 0.8705553335996804,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0756,
      "step": 2179
    },
    {
      "epoch": 0.87095485417499,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0804,
      "step": 2180
    },
    {
      "epoch": 0.8713543747502996,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0798,
      "step": 2181
    },
    {
      "epoch": 0.8717538953256093,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0764,
      "step": 2182
    },
    {
      "epoch": 0.8721534159009189,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0812,
      "step": 2183
    },
    {
      "epoch": 0.8725529364762286,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0809,
      "step": 2184
    },
    {
      "epoch": 0.8729524570515381,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.083,
      "step": 2185
    },
    {
      "epoch": 0.8733519776268478,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 2186
    },
    {
      "epoch": 0.8737514982021574,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0838,
      "step": 2187
    },
    {
      "epoch": 0.874151018777467,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0823,
      "step": 2188
    },
    {
      "epoch": 0.8745505393527767,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0783,
      "step": 2189
    },
    {
      "epoch": 0.8749500599280863,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0835,
      "step": 2190
    },
    {
      "epoch": 0.875349580503396,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0911,
      "step": 2191
    },
    {
      "epoch": 0.8757491010787055,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0837,
      "step": 2192
    },
    {
      "epoch": 0.8761486216540152,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0766,
      "step": 2193
    },
    {
      "epoch": 0.8765481422293248,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.069,
      "step": 2194
    },
    {
      "epoch": 0.8769476628046344,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0795,
      "step": 2195
    },
    {
      "epoch": 0.8773471833799441,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0834,
      "step": 2196
    },
    {
      "epoch": 0.8777467039552537,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0781,
      "step": 2197
    },
    {
      "epoch": 0.8781462245305633,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0975,
      "step": 2198
    },
    {
      "epoch": 0.8785457451058729,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0949,
      "step": 2199
    },
    {
      "epoch": 0.8789452656811826,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0892,
      "step": 2200
    },
    {
      "epoch": 0.8793447862564923,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0901,
      "step": 2201
    },
    {
      "epoch": 0.8797443068318018,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0891,
      "step": 2202
    },
    {
      "epoch": 0.8801438274071115,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0909,
      "step": 2203
    },
    {
      "epoch": 0.8805433479824211,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0848,
      "step": 2204
    },
    {
      "epoch": 0.8809428685577307,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.0754,
      "step": 2205
    },
    {
      "epoch": 0.8813423891330403,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0925,
      "step": 2206
    },
    {
      "epoch": 0.88174190970835,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0005,
      "loss": 1.0848,
      "step": 2207
    },
    {
      "epoch": 0.8821414302836597,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0901,
      "step": 2208
    },
    {
      "epoch": 0.8825409508589692,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0827,
      "step": 2209
    },
    {
      "epoch": 0.8829404714342789,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.086,
      "step": 2210
    },
    {
      "epoch": 0.8833399920095885,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0905,
      "step": 2211
    },
    {
      "epoch": 0.8837395125848981,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0909,
      "step": 2212
    },
    {
      "epoch": 0.8841390331602077,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0734,
      "step": 2213
    },
    {
      "epoch": 0.8845385537355174,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0825,
      "step": 2214
    },
    {
      "epoch": 0.884938074310827,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0819,
      "step": 2215
    },
    {
      "epoch": 0.8853375948861366,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0816,
      "step": 2216
    },
    {
      "epoch": 0.8857371154614463,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.077,
      "step": 2217
    },
    {
      "epoch": 0.8861366360367559,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0821,
      "step": 2218
    },
    {
      "epoch": 0.8865361566120655,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0697,
      "step": 2219
    },
    {
      "epoch": 0.8869356771873752,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.092,
      "step": 2220
    },
    {
      "epoch": 0.8873351977626848,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0843,
      "step": 2221
    },
    {
      "epoch": 0.8877347183379944,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0839,
      "step": 2222
    },
    {
      "epoch": 0.888134238913304,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0757,
      "step": 2223
    },
    {
      "epoch": 0.8885337594886137,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0729,
      "step": 2224
    },
    {
      "epoch": 0.8889332800639232,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0889,
      "step": 2225
    },
    {
      "epoch": 0.8893328006392329,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 2226
    },
    {
      "epoch": 0.8897323212145426,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0817,
      "step": 2227
    },
    {
      "epoch": 0.8901318417898522,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0808,
      "step": 2228
    },
    {
      "epoch": 0.8905313623651618,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0005,
      "loss": 1.083,
      "step": 2229
    },
    {
      "epoch": 0.8909308829404714,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0806,
      "step": 2230
    },
    {
      "epoch": 0.8913304035157811,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.0815,
      "step": 2231
    },
    {
      "epoch": 0.8917299240910906,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0864,
      "step": 2232
    },
    {
      "epoch": 0.8921294446664003,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0005,
      "loss": 1.0817,
      "step": 2233
    },
    {
      "epoch": 0.89252896524171,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0899,
      "step": 2234
    },
    {
      "epoch": 0.8929284858170196,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0895,
      "step": 2235
    },
    {
      "epoch": 0.8933280063923292,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0005,
      "loss": 1.0885,
      "step": 2236
    },
    {
      "epoch": 0.8937275269676388,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0829,
      "step": 2237
    },
    {
      "epoch": 0.8941270475429485,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0825,
      "step": 2238
    },
    {
      "epoch": 0.894526568118258,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.076,
      "step": 2239
    },
    {
      "epoch": 0.8949260886935677,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0869,
      "step": 2240
    },
    {
      "epoch": 0.8953256092688774,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.1001,
      "step": 2241
    },
    {
      "epoch": 0.895725129844187,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0898,
      "step": 2242
    },
    {
      "epoch": 0.8961246504194966,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0826,
      "step": 2243
    },
    {
      "epoch": 0.8965241709948062,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0732,
      "step": 2244
    },
    {
      "epoch": 0.8969236915701159,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0794,
      "step": 2245
    },
    {
      "epoch": 0.8973232121454255,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.08,
      "step": 2246
    },
    {
      "epoch": 0.8977227327207351,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0808,
      "step": 2247
    },
    {
      "epoch": 0.8981222532960448,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0863,
      "step": 2248
    },
    {
      "epoch": 0.8985217738713543,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0795,
      "step": 2249
    },
    {
      "epoch": 0.898921294446664,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0906,
      "step": 2250
    },
    {
      "epoch": 0.8993208150219736,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0877,
      "step": 2251
    },
    {
      "epoch": 0.8997203355972833,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0868,
      "step": 2252
    },
    {
      "epoch": 0.9001198561725929,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0811,
      "step": 2253
    },
    {
      "epoch": 0.9005193767479025,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0803,
      "step": 2254
    },
    {
      "epoch": 0.9009188973232122,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0768,
      "step": 2255
    },
    {
      "epoch": 0.9013184178985217,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.086,
      "step": 2256
    },
    {
      "epoch": 0.9017179384738314,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0769,
      "step": 2257
    },
    {
      "epoch": 0.902117459049141,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.1016,
      "step": 2258
    },
    {
      "epoch": 0.9025169796244507,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0872,
      "step": 2259
    },
    {
      "epoch": 0.9029165001997603,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0812,
      "step": 2260
    },
    {
      "epoch": 0.9033160207750699,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0794,
      "step": 2261
    },
    {
      "epoch": 0.9037155413503796,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0739,
      "step": 2262
    },
    {
      "epoch": 0.9041150619256891,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0857,
      "step": 2263
    },
    {
      "epoch": 0.9045145825009988,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0917,
      "step": 2264
    },
    {
      "epoch": 0.9049141030763085,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0746,
      "step": 2265
    },
    {
      "epoch": 0.905313623651618,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0783,
      "step": 2266
    },
    {
      "epoch": 0.9057131442269277,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0807,
      "step": 2267
    },
    {
      "epoch": 0.9061126648022373,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0884,
      "step": 2268
    },
    {
      "epoch": 0.906512185377547,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0951,
      "step": 2269
    },
    {
      "epoch": 0.9069117059528565,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0763,
      "step": 2270
    },
    {
      "epoch": 0.9073112265281662,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0804,
      "step": 2271
    },
    {
      "epoch": 0.9077107471034759,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0862,
      "step": 2272
    },
    {
      "epoch": 0.9081102676787854,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.08,
      "step": 2273
    },
    {
      "epoch": 0.9085097882540951,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0776,
      "step": 2274
    },
    {
      "epoch": 0.9089093088294047,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0809,
      "step": 2275
    },
    {
      "epoch": 0.9093088294047144,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.082,
      "step": 2276
    },
    {
      "epoch": 0.9097083499800239,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0815,
      "step": 2277
    },
    {
      "epoch": 0.9101078705553336,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0846,
      "step": 2278
    },
    {
      "epoch": 0.9105073911306433,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0794,
      "step": 2279
    },
    {
      "epoch": 0.9109069117059528,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.092,
      "step": 2280
    },
    {
      "epoch": 0.9113064322812625,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0829,
      "step": 2281
    },
    {
      "epoch": 0.9117059528565721,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0853,
      "step": 2282
    },
    {
      "epoch": 0.9121054734318818,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0817,
      "step": 2283
    },
    {
      "epoch": 0.9125049940071913,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0847,
      "step": 2284
    },
    {
      "epoch": 0.912904514582501,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0924,
      "step": 2285
    },
    {
      "epoch": 0.9133040351578107,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0895,
      "step": 2286
    },
    {
      "epoch": 0.9137035557331202,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005,
      "loss": 1.0839,
      "step": 2287
    },
    {
      "epoch": 0.9141030763084299,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0722,
      "step": 2288
    },
    {
      "epoch": 0.9145025968837395,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0005,
      "loss": 1.0778,
      "step": 2289
    },
    {
      "epoch": 0.9149021174590491,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005,
      "loss": 1.0716,
      "step": 2290
    },
    {
      "epoch": 0.9153016380343588,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0005,
      "loss": 1.0719,
      "step": 2291
    },
    {
      "epoch": 0.9157011586096684,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0005,
      "loss": 1.0893,
      "step": 2292
    },
    {
      "epoch": 0.9161006791849781,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0005,
      "loss": 1.0829,
      "step": 2293
    },
    {
      "epoch": 0.9165001997602876,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0005,
      "loss": 1.0919,
      "step": 2294
    },
    {
      "epoch": 0.9168997203355973,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0888,
      "step": 2295
    },
    {
      "epoch": 0.9172992409109069,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0005,
      "loss": 1.0839,
      "step": 2296
    },
    {
      "epoch": 0.9176987614862165,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 2297
    },
    {
      "epoch": 0.9180982820615262,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005,
      "loss": 1.0866,
      "step": 2298
    },
    {
      "epoch": 0.9184978026368358,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 2299
    },
    {
      "epoch": 0.9188973232121455,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0005,
      "loss": 1.0839,
      "step": 2300
    },
    {
      "epoch": 0.919296843787455,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.093,
      "step": 2301
    },
    {
      "epoch": 0.9196963643627647,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0956,
      "step": 2302
    },
    {
      "epoch": 0.9200958849380743,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0897,
      "step": 2303
    },
    {
      "epoch": 0.9204954055133839,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0885,
      "step": 2304
    },
    {
      "epoch": 0.9208949260886936,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0788,
      "step": 2305
    },
    {
      "epoch": 0.9212944466640032,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 2306
    },
    {
      "epoch": 0.9216939672393129,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.073,
      "step": 2307
    },
    {
      "epoch": 0.9220934878146224,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0877,
      "step": 2308
    },
    {
      "epoch": 0.9224930083899321,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0743,
      "step": 2309
    },
    {
      "epoch": 0.9228925289652418,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0828,
      "step": 2310
    },
    {
      "epoch": 0.9232920495405513,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 2311
    },
    {
      "epoch": 0.923691570115861,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 2312
    },
    {
      "epoch": 0.9240910906911706,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0901,
      "step": 2313
    },
    {
      "epoch": 0.9244906112664802,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0914,
      "step": 2314
    },
    {
      "epoch": 0.9248901318417898,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0767,
      "step": 2315
    },
    {
      "epoch": 0.9252896524170995,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0992,
      "step": 2316
    },
    {
      "epoch": 0.9256891729924092,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0761,
      "step": 2317
    },
    {
      "epoch": 0.9260886935677187,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0734,
      "step": 2318
    },
    {
      "epoch": 0.9264882141430284,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 2319
    },
    {
      "epoch": 0.926887734718338,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0879,
      "step": 2320
    },
    {
      "epoch": 0.9272872552936476,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0711,
      "step": 2321
    },
    {
      "epoch": 0.9276867758689572,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0931,
      "step": 2322
    },
    {
      "epoch": 0.9280862964442669,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.089,
      "step": 2323
    },
    {
      "epoch": 0.9284858170195766,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0841,
      "step": 2324
    },
    {
      "epoch": 0.9288853375948861,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0916,
      "step": 2325
    },
    {
      "epoch": 0.9292848581701958,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.075,
      "step": 2326
    },
    {
      "epoch": 0.9296843787455054,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0823,
      "step": 2327
    },
    {
      "epoch": 0.930083899320815,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0799,
      "step": 2328
    },
    {
      "epoch": 0.9304834198961246,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0797,
      "step": 2329
    },
    {
      "epoch": 0.9308829404714343,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.083,
      "step": 2330
    },
    {
      "epoch": 0.931282461046744,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0775,
      "step": 2331
    },
    {
      "epoch": 0.9316819816220535,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0888,
      "step": 2332
    },
    {
      "epoch": 0.9320815021973632,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0827,
      "step": 2333
    },
    {
      "epoch": 0.9324810227726728,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.081,
      "step": 2334
    },
    {
      "epoch": 0.9328805433479824,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0837,
      "step": 2335
    },
    {
      "epoch": 0.9332800639232921,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0791,
      "step": 2336
    },
    {
      "epoch": 0.9336795844986017,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0657,
      "step": 2337
    },
    {
      "epoch": 0.9340791050739113,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0882,
      "step": 2338
    },
    {
      "epoch": 0.9344786256492209,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 2339
    },
    {
      "epoch": 0.9348781462245306,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.089,
      "step": 2340
    },
    {
      "epoch": 0.9352776667998401,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.083,
      "step": 2341
    },
    {
      "epoch": 0.9356771873751498,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0711,
      "step": 2342
    },
    {
      "epoch": 0.9360767079504595,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0846,
      "step": 2343
    },
    {
      "epoch": 0.9364762285257691,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0885,
      "step": 2344
    },
    {
      "epoch": 0.9368757491010787,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0791,
      "step": 2345
    },
    {
      "epoch": 0.9372752696763883,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0849,
      "step": 2346
    },
    {
      "epoch": 0.937674790251698,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.082,
      "step": 2347
    },
    {
      "epoch": 0.9380743108270075,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0813,
      "step": 2348
    },
    {
      "epoch": 0.9384738314023172,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0853,
      "step": 2349
    },
    {
      "epoch": 0.9388733519776269,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0889,
      "step": 2350
    },
    {
      "epoch": 0.9392728725529365,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0935,
      "step": 2351
    },
    {
      "epoch": 0.9396723931282461,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.083,
      "step": 2352
    },
    {
      "epoch": 0.9400719137035557,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0813,
      "step": 2353
    },
    {
      "epoch": 0.9404714342788654,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0818,
      "step": 2354
    },
    {
      "epoch": 0.9408709548541749,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0722,
      "step": 2355
    },
    {
      "epoch": 0.9412704754294846,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0828,
      "step": 2356
    },
    {
      "epoch": 0.9416699960047943,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.082,
      "step": 2357
    },
    {
      "epoch": 0.9420695165801038,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.084,
      "step": 2358
    },
    {
      "epoch": 0.9424690371554135,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0929,
      "step": 2359
    },
    {
      "epoch": 0.9428685577307231,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0749,
      "step": 2360
    },
    {
      "epoch": 0.9432680783060328,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 2361
    },
    {
      "epoch": 0.9436675988813424,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0817,
      "step": 2362
    },
    {
      "epoch": 0.944067119456652,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0852,
      "step": 2363
    },
    {
      "epoch": 0.9444666400319617,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0837,
      "step": 2364
    },
    {
      "epoch": 0.9448661606072712,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.094,
      "step": 2365
    },
    {
      "epoch": 0.9452656811825809,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.085,
      "step": 2366
    },
    {
      "epoch": 0.9456652017578905,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.074,
      "step": 2367
    },
    {
      "epoch": 0.9460647223332002,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0822,
      "step": 2368
    },
    {
      "epoch": 0.9464642429085098,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0853,
      "step": 2369
    },
    {
      "epoch": 0.9468637634838194,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.082,
      "step": 2370
    },
    {
      "epoch": 0.9472632840591291,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0746,
      "step": 2371
    },
    {
      "epoch": 0.9476628046344386,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0867,
      "step": 2372
    },
    {
      "epoch": 0.9480623252097483,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0766,
      "step": 2373
    },
    {
      "epoch": 0.9484618457850579,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0766,
      "step": 2374
    },
    {
      "epoch": 0.9488613663603676,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0766,
      "step": 2375
    },
    {
      "epoch": 0.9492608869356772,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.079,
      "step": 2376
    },
    {
      "epoch": 0.9496604075109868,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0774,
      "step": 2377
    },
    {
      "epoch": 0.9500599280862965,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0767,
      "step": 2378
    },
    {
      "epoch": 0.950459448661606,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 2379
    },
    {
      "epoch": 0.9508589692369157,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 2380
    },
    {
      "epoch": 0.9512584898122254,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 2381
    },
    {
      "epoch": 0.951658010387535,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0774,
      "step": 2382
    },
    {
      "epoch": 0.9520575309628446,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.076,
      "step": 2383
    },
    {
      "epoch": 0.9524570515381542,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0772,
      "step": 2384
    },
    {
      "epoch": 0.9528565721134639,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 2385
    },
    {
      "epoch": 0.9532560926887734,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0837,
      "step": 2386
    },
    {
      "epoch": 0.9536556132640831,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 2387
    },
    {
      "epoch": 0.9540551338393928,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0914,
      "step": 2388
    },
    {
      "epoch": 0.9544546544147023,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 2389
    },
    {
      "epoch": 0.954854174990012,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0949,
      "step": 2390
    },
    {
      "epoch": 0.9552536955653216,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 2391
    },
    {
      "epoch": 0.9556532161406313,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0875,
      "step": 2392
    },
    {
      "epoch": 0.9560527367159408,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0765,
      "step": 2393
    },
    {
      "epoch": 0.9564522572912505,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.082,
      "step": 2394
    },
    {
      "epoch": 0.9568517778665602,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0874,
      "step": 2395
    },
    {
      "epoch": 0.9572512984418697,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 2396
    },
    {
      "epoch": 0.9576508190171794,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 2397
    },
    {
      "epoch": 0.958050339592489,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0975,
      "step": 2398
    },
    {
      "epoch": 0.9584498601677987,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0845,
      "step": 2399
    },
    {
      "epoch": 0.9588493807431082,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0842,
      "step": 2400
    },
    {
      "epoch": 0.9592489013184179,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0803,
      "step": 2401
    },
    {
      "epoch": 0.9596484218937276,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0792,
      "step": 2402
    },
    {
      "epoch": 0.9600479424690371,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0803,
      "step": 2403
    },
    {
      "epoch": 0.9604474630443468,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0873,
      "step": 2404
    },
    {
      "epoch": 0.9608469836196564,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0868,
      "step": 2405
    },
    {
      "epoch": 0.961246504194966,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0779,
      "step": 2406
    },
    {
      "epoch": 0.9616460247702757,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0799,
      "step": 2407
    },
    {
      "epoch": 0.9620455453455853,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 2408
    },
    {
      "epoch": 0.962445065920895,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005,
      "loss": 1.0767,
      "step": 2409
    },
    {
      "epoch": 0.9628445864962045,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0799,
      "step": 2410
    },
    {
      "epoch": 0.9632441070715142,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0005,
      "loss": 1.081,
      "step": 2411
    },
    {
      "epoch": 0.9636436276468238,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0807,
      "step": 2412
    },
    {
      "epoch": 0.9640431482221334,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.0921,
      "step": 2413
    },
    {
      "epoch": 0.9644426687974431,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0005,
      "loss": 1.0741,
      "step": 2414
    },
    {
      "epoch": 0.9648421893727527,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0791,
      "step": 2415
    },
    {
      "epoch": 0.9652417099480624,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0826,
      "step": 2416
    },
    {
      "epoch": 0.9656412305233719,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0807,
      "step": 2417
    },
    {
      "epoch": 0.9660407510986816,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0904,
      "step": 2418
    },
    {
      "epoch": 0.9664402716739912,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 2419
    },
    {
      "epoch": 0.9668397922493008,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0807,
      "step": 2420
    },
    {
      "epoch": 0.9672393128246105,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0831,
      "step": 2421
    },
    {
      "epoch": 0.9676388333999201,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0857,
      "step": 2422
    },
    {
      "epoch": 0.9680383539752297,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0005,
      "loss": 1.0795,
      "step": 2423
    },
    {
      "epoch": 0.9684378745505393,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 2424
    },
    {
      "epoch": 0.968837395125849,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.1012,
      "step": 2425
    },
    {
      "epoch": 0.9692369157011587,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 2426
    },
    {
      "epoch": 0.9696364362764682,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0868,
      "step": 2427
    },
    {
      "epoch": 0.9700359568517779,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0005,
      "loss": 1.0822,
      "step": 2428
    },
    {
      "epoch": 0.9704354774270875,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 2429
    },
    {
      "epoch": 0.9708349980023971,
      "grad_norm": 0.28125,
      "learning_rate": 0.0005,
      "loss": 1.0805,
      "step": 2430
    },
    {
      "epoch": 0.9712345185777067,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.0773,
      "step": 2431
    },
    {
      "epoch": 0.9716340391530164,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0818,
      "step": 2432
    },
    {
      "epoch": 0.9720335597283261,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0005,
      "loss": 1.0846,
      "step": 2433
    },
    {
      "epoch": 0.9724330803036356,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0744,
      "step": 2434
    },
    {
      "epoch": 0.9728326008789453,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.079,
      "step": 2435
    },
    {
      "epoch": 0.9732321214542549,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0005,
      "loss": 1.0868,
      "step": 2436
    },
    {
      "epoch": 0.9736316420295645,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.075,
      "step": 2437
    },
    {
      "epoch": 0.9740311626048741,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.0826,
      "step": 2438
    },
    {
      "epoch": 0.9744306831801838,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0786,
      "step": 2439
    },
    {
      "epoch": 0.9748302037554935,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0005,
      "loss": 1.0881,
      "step": 2440
    },
    {
      "epoch": 0.975229724330803,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 2441
    },
    {
      "epoch": 0.9756292449061127,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0892,
      "step": 2442
    },
    {
      "epoch": 0.9760287654814223,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0711,
      "step": 2443
    },
    {
      "epoch": 0.9764282860567319,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.083,
      "step": 2444
    },
    {
      "epoch": 0.9768278066320415,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0847,
      "step": 2445
    },
    {
      "epoch": 0.9772273272073512,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0824,
      "step": 2446
    },
    {
      "epoch": 0.9776268477826608,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0818,
      "step": 2447
    },
    {
      "epoch": 0.9780263683579704,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0764,
      "step": 2448
    },
    {
      "epoch": 0.9784258889332801,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0787,
      "step": 2449
    },
    {
      "epoch": 0.9788254095085897,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 2450
    },
    {
      "epoch": 0.9792249300838993,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0813,
      "step": 2451
    },
    {
      "epoch": 0.979624450659209,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0777,
      "step": 2452
    },
    {
      "epoch": 0.9800239712345186,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 2453
    },
    {
      "epoch": 0.9804234918098282,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0846,
      "step": 2454
    },
    {
      "epoch": 0.9808230123851378,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0774,
      "step": 2455
    },
    {
      "epoch": 0.9812225329604475,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 2456
    },
    {
      "epoch": 0.981622053535757,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0672,
      "step": 2457
    },
    {
      "epoch": 0.9820215741110667,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0734,
      "step": 2458
    },
    {
      "epoch": 0.9824210946863764,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0762,
      "step": 2459
    },
    {
      "epoch": 0.982820615261686,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.073,
      "step": 2460
    },
    {
      "epoch": 0.9832201358369956,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0781,
      "step": 2461
    },
    {
      "epoch": 0.9836196564123052,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0768,
      "step": 2462
    },
    {
      "epoch": 0.9840191769876149,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0938,
      "step": 2463
    },
    {
      "epoch": 0.9844186975629244,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 2464
    },
    {
      "epoch": 0.9848182181382341,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0898,
      "step": 2465
    },
    {
      "epoch": 0.9852177387135438,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0787,
      "step": 2466
    },
    {
      "epoch": 0.9856172592888534,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0773,
      "step": 2467
    },
    {
      "epoch": 0.986016779864163,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.09,
      "step": 2468
    },
    {
      "epoch": 0.9864163004394726,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 2469
    },
    {
      "epoch": 0.9868158210147823,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0769,
      "step": 2470
    },
    {
      "epoch": 0.9872153415900919,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0722,
      "step": 2471
    },
    {
      "epoch": 0.9876148621654015,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0826,
      "step": 2472
    },
    {
      "epoch": 0.9880143827407112,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0771,
      "step": 2473
    },
    {
      "epoch": 0.9884139033160207,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.091,
      "step": 2474
    },
    {
      "epoch": 0.9888134238913304,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0713,
      "step": 2475
    },
    {
      "epoch": 0.98921294446664,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0769,
      "step": 2476
    },
    {
      "epoch": 0.9896124650419497,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0755,
      "step": 2477
    },
    {
      "epoch": 0.9900119856172593,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0781,
      "step": 2478
    },
    {
      "epoch": 0.9904115061925689,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 2479
    },
    {
      "epoch": 0.9908110267678786,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0894,
      "step": 2480
    },
    {
      "epoch": 0.9912105473431881,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0801,
      "step": 2481
    },
    {
      "epoch": 0.9916100679184978,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0794,
      "step": 2482
    },
    {
      "epoch": 0.9920095884938074,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 2483
    },
    {
      "epoch": 0.9924091090691171,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0839,
      "step": 2484
    },
    {
      "epoch": 0.9928086296444267,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.079,
      "step": 2485
    },
    {
      "epoch": 0.9932081502197363,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0948,
      "step": 2486
    },
    {
      "epoch": 0.993607670795046,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0843,
      "step": 2487
    },
    {
      "epoch": 0.9940071913703555,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 2488
    },
    {
      "epoch": 0.9944067119456652,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0791,
      "step": 2489
    },
    {
      "epoch": 0.9948062325209748,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0864,
      "step": 2490
    },
    {
      "epoch": 0.9952057530962845,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0937,
      "step": 2491
    },
    {
      "epoch": 0.9956052736715941,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0901,
      "step": 2492
    },
    {
      "epoch": 0.9960047942469037,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0802,
      "step": 2493
    },
    {
      "epoch": 0.9964043148222134,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0872,
      "step": 2494
    },
    {
      "epoch": 0.9968038353975229,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0796,
      "step": 2495
    },
    {
      "epoch": 0.9972033559728326,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0826,
      "step": 2496
    },
    {
      "epoch": 0.9976028765481423,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 2497
    },
    {
      "epoch": 0.9980023971234518,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0789,
      "step": 2498
    },
    {
      "epoch": 0.9984019176987615,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0756,
      "step": 2499
    },
    {
      "epoch": 0.9988014382740711,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0731,
      "step": 2500
    },
    {
      "epoch": 0.9992009588493808,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0766,
      "step": 2501
    },
    {
      "epoch": 0.9996004794246903,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0738,
      "step": 2502
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0765,
      "step": 2503
    },
    {
      "epoch": 1.0003995205753096,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0782,
      "step": 2504
    },
    {
      "epoch": 1.0007990411506194,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0832,
      "step": 2505
    },
    {
      "epoch": 1.001198561725929,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0861,
      "step": 2506
    },
    {
      "epoch": 1.0015980823012385,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0797,
      "step": 2507
    },
    {
      "epoch": 1.001997602876548,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0775,
      "step": 2508
    },
    {
      "epoch": 1.0023971234518578,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0867,
      "step": 2509
    },
    {
      "epoch": 1.0027966440271674,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0846,
      "step": 2510
    },
    {
      "epoch": 1.003196164602477,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0874,
      "step": 2511
    },
    {
      "epoch": 1.0035956851777867,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0757,
      "step": 2512
    },
    {
      "epoch": 1.0039952057530963,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.078,
      "step": 2513
    },
    {
      "epoch": 1.0043947263284059,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0744,
      "step": 2514
    },
    {
      "epoch": 1.0047942469037154,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0845,
      "step": 2515
    },
    {
      "epoch": 1.0051937674790252,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0862,
      "step": 2516
    },
    {
      "epoch": 1.0055932880543348,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0784,
      "step": 2517
    },
    {
      "epoch": 1.0059928086296444,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0783,
      "step": 2518
    },
    {
      "epoch": 1.0063923292049541,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0889,
      "step": 2519
    },
    {
      "epoch": 1.0067918497802637,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 2520
    },
    {
      "epoch": 1.0071913703555733,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0771,
      "step": 2521
    },
    {
      "epoch": 1.0075908909308828,
      "grad_norm": 0.265625,
      "learning_rate": 0.0005,
      "loss": 1.089,
      "step": 2522
    },
    {
      "epoch": 1.0079904115061926,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0737,
      "step": 2523
    },
    {
      "epoch": 1.0083899320815022,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0005,
      "loss": 1.0848,
      "step": 2524
    },
    {
      "epoch": 1.0087894526568117,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 2525
    },
    {
      "epoch": 1.0091889732321215,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0766,
      "step": 2526
    },
    {
      "epoch": 1.009588493807431,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 2527
    },
    {
      "epoch": 1.0099880143827407,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0005,
      "loss": 1.084,
      "step": 2528
    },
    {
      "epoch": 1.0103875349580504,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0789,
      "step": 2529
    },
    {
      "epoch": 1.01078705553336,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 2530
    },
    {
      "epoch": 1.0111865761086696,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.07,
      "step": 2531
    },
    {
      "epoch": 1.0115860966839791,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0802,
      "step": 2532
    },
    {
      "epoch": 1.011985617259289,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0714,
      "step": 2533
    },
    {
      "epoch": 1.0123851378345985,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.1004,
      "step": 2534
    },
    {
      "epoch": 1.012784658409908,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 2535
    },
    {
      "epoch": 1.0131841789852178,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0879,
      "step": 2536
    },
    {
      "epoch": 1.0135836995605274,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 2537
    },
    {
      "epoch": 1.013983220135837,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 2538
    },
    {
      "epoch": 1.0143827407111465,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 2539
    },
    {
      "epoch": 1.0147822612864563,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0889,
      "step": 2540
    },
    {
      "epoch": 1.0151817818617659,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 2541
    },
    {
      "epoch": 1.0155813024370755,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 2542
    },
    {
      "epoch": 1.0159808230123852,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0749,
      "step": 2543
    },
    {
      "epoch": 1.0163803435876948,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0789,
      "step": 2544
    },
    {
      "epoch": 1.0167798641630044,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.078,
      "step": 2545
    },
    {
      "epoch": 1.017179384738314,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0745,
      "step": 2546
    },
    {
      "epoch": 1.0175789053136237,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0716,
      "step": 2547
    },
    {
      "epoch": 1.0179784258889333,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0704,
      "step": 2548
    },
    {
      "epoch": 1.0183779464642428,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0754,
      "step": 2549
    },
    {
      "epoch": 1.0187774670395526,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0799,
      "step": 2550
    },
    {
      "epoch": 1.0191769876148622,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0722,
      "step": 2551
    },
    {
      "epoch": 1.0195765081901718,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 2552
    },
    {
      "epoch": 1.0199760287654813,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0734,
      "step": 2553
    },
    {
      "epoch": 1.020375549340791,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0901,
      "step": 2554
    },
    {
      "epoch": 1.0207750699161007,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0706,
      "step": 2555
    },
    {
      "epoch": 1.0211745904914102,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 2556
    },
    {
      "epoch": 1.02157411106672,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 2557
    },
    {
      "epoch": 1.0219736316420296,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0734,
      "step": 2558
    },
    {
      "epoch": 1.0223731522173392,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.085,
      "step": 2559
    },
    {
      "epoch": 1.0227726727926487,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.088,
      "step": 2560
    },
    {
      "epoch": 1.0231721933679585,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0746,
      "step": 2561
    },
    {
      "epoch": 1.023571713943268,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0799,
      "step": 2562
    },
    {
      "epoch": 1.0239712345185776,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0911,
      "step": 2563
    },
    {
      "epoch": 1.0243707550938874,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0812,
      "step": 2564
    },
    {
      "epoch": 1.024770275669197,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0805,
      "step": 2565
    },
    {
      "epoch": 1.0251697962445065,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 2566
    },
    {
      "epoch": 1.0255693168198161,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0717,
      "step": 2567
    },
    {
      "epoch": 1.025968837395126,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 2568
    },
    {
      "epoch": 1.0263683579704355,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0802,
      "step": 2569
    },
    {
      "epoch": 1.026767878545745,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0815,
      "step": 2570
    },
    {
      "epoch": 1.0271673991210548,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0735,
      "step": 2571
    },
    {
      "epoch": 1.0275669196963644,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 2572
    },
    {
      "epoch": 1.027966440271674,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0914,
      "step": 2573
    },
    {
      "epoch": 1.0283659608469837,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0865,
      "step": 2574
    },
    {
      "epoch": 1.0287654814222933,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0772,
      "step": 2575
    },
    {
      "epoch": 1.0291650019976029,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0816,
      "step": 2576
    },
    {
      "epoch": 1.0295645225729124,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0834,
      "step": 2577
    },
    {
      "epoch": 1.0299640431482222,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 2578
    },
    {
      "epoch": 1.0303635637235318,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0814,
      "step": 2579
    },
    {
      "epoch": 1.0307630842988413,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 2580
    },
    {
      "epoch": 1.0311626048741511,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0877,
      "step": 2581
    },
    {
      "epoch": 1.0315621254494607,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0818,
      "step": 2582
    },
    {
      "epoch": 1.0319616460247703,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 2583
    },
    {
      "epoch": 1.0323611666000798,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 2584
    },
    {
      "epoch": 1.0327606871753896,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0797,
      "step": 2585
    },
    {
      "epoch": 1.0331602077506992,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0776,
      "step": 2586
    },
    {
      "epoch": 1.0335597283260087,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0821,
      "step": 2587
    },
    {
      "epoch": 1.0339592489013185,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0804,
      "step": 2588
    },
    {
      "epoch": 1.034358769476628,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.0765,
      "step": 2589
    },
    {
      "epoch": 1.0347582900519376,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0889,
      "step": 2590
    },
    {
      "epoch": 1.0351578106272472,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.079,
      "step": 2591
    },
    {
      "epoch": 1.035557331202557,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0696,
      "step": 2592
    },
    {
      "epoch": 1.0359568517778666,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 2593
    },
    {
      "epoch": 1.0363563723531761,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 2594
    },
    {
      "epoch": 1.036755892928486,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0768,
      "step": 2595
    },
    {
      "epoch": 1.0371554135037955,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0877,
      "step": 2596
    },
    {
      "epoch": 1.037554934079105,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 2597
    },
    {
      "epoch": 1.0379544546544146,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 2598
    },
    {
      "epoch": 1.0383539752297244,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0883,
      "step": 2599
    },
    {
      "epoch": 1.038753495805034,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0764,
      "step": 2600
    },
    {
      "epoch": 1.0391530163803435,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 2601
    },
    {
      "epoch": 1.0395525369556533,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0753,
      "step": 2602
    },
    {
      "epoch": 1.0399520575309629,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 2603
    },
    {
      "epoch": 1.0403515781062724,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0729,
      "step": 2604
    },
    {
      "epoch": 1.040751098681582,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 2605
    },
    {
      "epoch": 1.0411506192568918,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 2606
    },
    {
      "epoch": 1.0415501398322013,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 2607
    },
    {
      "epoch": 1.041949660407511,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0809,
      "step": 2608
    },
    {
      "epoch": 1.0423491809828207,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0774,
      "step": 2609
    },
    {
      "epoch": 1.0427487015581303,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0767,
      "step": 2610
    },
    {
      "epoch": 1.0431482221334398,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0802,
      "step": 2611
    },
    {
      "epoch": 1.0435477427087494,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0809,
      "step": 2612
    },
    {
      "epoch": 1.0439472632840592,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 2613
    },
    {
      "epoch": 1.0443467838593687,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0818,
      "step": 2614
    },
    {
      "epoch": 1.0447463044346783,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0757,
      "step": 2615
    },
    {
      "epoch": 1.045145825009988,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0888,
      "step": 2616
    },
    {
      "epoch": 1.0455453455852977,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 2617
    },
    {
      "epoch": 1.0459448661606072,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0732,
      "step": 2618
    },
    {
      "epoch": 1.046344386735917,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0784,
      "step": 2619
    },
    {
      "epoch": 1.0467439073112266,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0753,
      "step": 2620
    },
    {
      "epoch": 1.0471434278865361,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 2621
    },
    {
      "epoch": 1.0475429484618457,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0818,
      "step": 2622
    },
    {
      "epoch": 1.0479424690371555,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0712,
      "step": 2623
    },
    {
      "epoch": 1.048341989612465,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.076,
      "step": 2624
    },
    {
      "epoch": 1.0487415101877746,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 2625
    },
    {
      "epoch": 1.0491410307630844,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 2626
    },
    {
      "epoch": 1.049540551338394,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 2627
    },
    {
      "epoch": 1.0499400719137035,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0752,
      "step": 2628
    },
    {
      "epoch": 1.050339592489013,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0808,
      "step": 2629
    },
    {
      "epoch": 1.0507391130643229,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0859,
      "step": 2630
    },
    {
      "epoch": 1.0511386336396324,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.078,
      "step": 2631
    },
    {
      "epoch": 1.051538154214942,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0727,
      "step": 2632
    },
    {
      "epoch": 1.0519376747902518,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0862,
      "step": 2633
    },
    {
      "epoch": 1.0523371953655614,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0807,
      "step": 2634
    },
    {
      "epoch": 1.052736715940871,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0816,
      "step": 2635
    },
    {
      "epoch": 1.0531362365161805,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 2636
    },
    {
      "epoch": 1.0535357570914903,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.085,
      "step": 2637
    },
    {
      "epoch": 1.0539352776667998,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0005,
      "loss": 1.0825,
      "step": 2638
    },
    {
      "epoch": 1.0543347982421094,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0829,
      "step": 2639
    },
    {
      "epoch": 1.0547343188174192,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0811,
      "step": 2640
    },
    {
      "epoch": 1.0551338393927288,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0721,
      "step": 2641
    },
    {
      "epoch": 1.0555333599680383,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0711,
      "step": 2642
    },
    {
      "epoch": 1.0559328805433479,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 2643
    },
    {
      "epoch": 1.0563324011186577,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.0777,
      "step": 2644
    },
    {
      "epoch": 1.0567319216939672,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0737,
      "step": 2645
    },
    {
      "epoch": 1.0571314422692768,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0005,
      "loss": 1.0696,
      "step": 2646
    },
    {
      "epoch": 1.0575309628445866,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0741,
      "step": 2647
    },
    {
      "epoch": 1.0579304834198962,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0741,
      "step": 2648
    },
    {
      "epoch": 1.0583300039952057,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0787,
      "step": 2649
    },
    {
      "epoch": 1.0587295245705153,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0805,
      "step": 2650
    },
    {
      "epoch": 1.059129045145825,
      "grad_norm": 0.203125,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 2651
    },
    {
      "epoch": 1.0595285657211346,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0826,
      "step": 2652
    },
    {
      "epoch": 1.0599280862964442,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 2653
    },
    {
      "epoch": 1.060327606871754,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0813,
      "step": 2654
    },
    {
      "epoch": 1.0607271274470635,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0812,
      "step": 2655
    },
    {
      "epoch": 1.061126648022373,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.077,
      "step": 2656
    },
    {
      "epoch": 1.0615261685976827,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0859,
      "step": 2657
    },
    {
      "epoch": 1.0619256891729925,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0808,
      "step": 2658
    },
    {
      "epoch": 1.062325209748302,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0822,
      "step": 2659
    },
    {
      "epoch": 1.0627247303236116,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 2660
    },
    {
      "epoch": 1.0631242508989214,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0843,
      "step": 2661
    },
    {
      "epoch": 1.063523771474231,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 2662
    },
    {
      "epoch": 1.0639232920495405,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0749,
      "step": 2663
    },
    {
      "epoch": 1.0643228126248503,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0746,
      "step": 2664
    },
    {
      "epoch": 1.0647223332001599,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0822,
      "step": 2665
    },
    {
      "epoch": 1.0651218537754694,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0793,
      "step": 2666
    },
    {
      "epoch": 1.065521374350779,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0784,
      "step": 2667
    },
    {
      "epoch": 1.0659208949260888,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 2668
    },
    {
      "epoch": 1.0663204155013983,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 2669
    },
    {
      "epoch": 1.066719936076708,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0756,
      "step": 2670
    },
    {
      "epoch": 1.0671194566520175,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 2671
    },
    {
      "epoch": 1.0675189772273272,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0769,
      "step": 2672
    },
    {
      "epoch": 1.0679184978026368,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.082,
      "step": 2673
    },
    {
      "epoch": 1.0683180183779464,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 2674
    },
    {
      "epoch": 1.0687175389532562,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0792,
      "step": 2675
    },
    {
      "epoch": 1.0691170595285657,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0849,
      "step": 2676
    },
    {
      "epoch": 1.0695165801038753,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0808,
      "step": 2677
    },
    {
      "epoch": 1.069916100679185,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 2678
    },
    {
      "epoch": 1.0703156212544946,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 2679
    },
    {
      "epoch": 1.0707151418298042,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0816,
      "step": 2680
    },
    {
      "epoch": 1.0711146624051138,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0811,
      "step": 2681
    },
    {
      "epoch": 1.0715141829804236,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0665,
      "step": 2682
    },
    {
      "epoch": 1.0719137035557331,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0737,
      "step": 2683
    },
    {
      "epoch": 1.0723132241310427,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0866,
      "step": 2684
    },
    {
      "epoch": 1.0727127447063525,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 2685
    },
    {
      "epoch": 1.073112265281662,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0761,
      "step": 2686
    },
    {
      "epoch": 1.0735117858569716,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 2687
    },
    {
      "epoch": 1.0739113064322812,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0717,
      "step": 2688
    },
    {
      "epoch": 1.074310827007591,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 2689
    },
    {
      "epoch": 1.0747103475829005,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0765,
      "step": 2690
    },
    {
      "epoch": 1.07510986815821,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 2691
    },
    {
      "epoch": 1.0755093887335199,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 2692
    },
    {
      "epoch": 1.0759089093088294,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0742,
      "step": 2693
    },
    {
      "epoch": 1.076308429884139,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0822,
      "step": 2694
    },
    {
      "epoch": 1.0767079504594486,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0765,
      "step": 2695
    },
    {
      "epoch": 1.0771074710347583,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0795,
      "step": 2696
    },
    {
      "epoch": 1.077506991610068,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0787,
      "step": 2697
    },
    {
      "epoch": 1.0779065121853775,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0836,
      "step": 2698
    },
    {
      "epoch": 1.0783060327606873,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 2699
    },
    {
      "epoch": 1.0787055533359968,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0872,
      "step": 2700
    },
    {
      "epoch": 1.0791050739113064,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 2701
    },
    {
      "epoch": 1.079504594486616,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 2702
    },
    {
      "epoch": 1.0799041150619257,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0848,
      "step": 2703
    },
    {
      "epoch": 1.0803036356372353,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0764,
      "step": 2704
    },
    {
      "epoch": 1.0807031562125449,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0835,
      "step": 2705
    },
    {
      "epoch": 1.0811026767878547,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.0786,
      "step": 2706
    },
    {
      "epoch": 1.0815021973631642,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 2707
    },
    {
      "epoch": 1.0819017179384738,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.074,
      "step": 2708
    },
    {
      "epoch": 1.0823012385137836,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0792,
      "step": 2709
    },
    {
      "epoch": 1.0827007590890931,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0825,
      "step": 2710
    },
    {
      "epoch": 1.0831002796644027,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0741,
      "step": 2711
    },
    {
      "epoch": 1.0834998002397123,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 2712
    },
    {
      "epoch": 1.083899320815022,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0727,
      "step": 2713
    },
    {
      "epoch": 1.0842988413903316,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0811,
      "step": 2714
    },
    {
      "epoch": 1.0846983619656412,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0869,
      "step": 2715
    },
    {
      "epoch": 1.0850978825409507,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 2716
    },
    {
      "epoch": 1.0854974031162605,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0824,
      "step": 2717
    },
    {
      "epoch": 1.08589692369157,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0882,
      "step": 2718
    },
    {
      "epoch": 1.0862964442668797,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0773,
      "step": 2719
    },
    {
      "epoch": 1.0866959648421894,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0885,
      "step": 2720
    },
    {
      "epoch": 1.087095485417499,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0795,
      "step": 2721
    },
    {
      "epoch": 1.0874950059928086,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 2722
    },
    {
      "epoch": 1.0878945265681184,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 2723
    },
    {
      "epoch": 1.088294047143428,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 2724
    },
    {
      "epoch": 1.0886935677187375,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0754,
      "step": 2725
    },
    {
      "epoch": 1.089093088294047,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0727,
      "step": 2726
    },
    {
      "epoch": 1.0894926088693568,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0838,
      "step": 2727
    },
    {
      "epoch": 1.0898921294446664,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0812,
      "step": 2728
    },
    {
      "epoch": 1.090291650019976,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0764,
      "step": 2729
    },
    {
      "epoch": 1.0906911705952858,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0813,
      "step": 2730
    },
    {
      "epoch": 1.0910906911705953,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 2731
    },
    {
      "epoch": 1.0914902117459049,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.079,
      "step": 2732
    },
    {
      "epoch": 1.0918897323212144,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0788,
      "step": 2733
    },
    {
      "epoch": 1.0922892528965242,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0786,
      "step": 2734
    },
    {
      "epoch": 1.0926887734718338,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0782,
      "step": 2735
    },
    {
      "epoch": 1.0930882940471434,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.09,
      "step": 2736
    },
    {
      "epoch": 1.0934878146224531,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0812,
      "step": 2737
    },
    {
      "epoch": 1.0938873351977627,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0774,
      "step": 2738
    },
    {
      "epoch": 1.0942868557730723,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 2739
    },
    {
      "epoch": 1.0946863763483818,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.08,
      "step": 2740
    },
    {
      "epoch": 1.0950858969236916,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0704,
      "step": 2741
    },
    {
      "epoch": 1.0954854174990012,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0779,
      "step": 2742
    },
    {
      "epoch": 1.0958849380743108,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0909,
      "step": 2743
    },
    {
      "epoch": 1.0962844586496205,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0776,
      "step": 2744
    },
    {
      "epoch": 1.09668397922493,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 2745
    },
    {
      "epoch": 1.0970834998002397,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 2746
    },
    {
      "epoch": 1.0974830203755492,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0757,
      "step": 2747
    },
    {
      "epoch": 1.097882540950859,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0745,
      "step": 2748
    },
    {
      "epoch": 1.0982820615261686,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0828,
      "step": 2749
    },
    {
      "epoch": 1.0986815821014781,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0805,
      "step": 2750
    },
    {
      "epoch": 1.099081102676788,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 2751
    },
    {
      "epoch": 1.0994806232520975,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0752,
      "step": 2752
    },
    {
      "epoch": 1.099880143827407,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 2753
    },
    {
      "epoch": 1.1002796644027169,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0787,
      "step": 2754
    },
    {
      "epoch": 1.1006791849780264,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 2755
    },
    {
      "epoch": 1.101078705553336,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0651,
      "step": 2756
    },
    {
      "epoch": 1.1014782261286455,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.088,
      "step": 2757
    },
    {
      "epoch": 1.1018777467039553,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0832,
      "step": 2758
    },
    {
      "epoch": 1.102277267279265,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0817,
      "step": 2759
    },
    {
      "epoch": 1.1026767878545745,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0777,
      "step": 2760
    },
    {
      "epoch": 1.103076308429884,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 2761
    },
    {
      "epoch": 1.1034758290051938,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0875,
      "step": 2762
    },
    {
      "epoch": 1.1038753495805034,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 2763
    },
    {
      "epoch": 1.104274870155813,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0728,
      "step": 2764
    },
    {
      "epoch": 1.1046743907311227,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 2765
    },
    {
      "epoch": 1.1050739113064323,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 2766
    },
    {
      "epoch": 1.1054734318817419,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0835,
      "step": 2767
    },
    {
      "epoch": 1.1058729524570516,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 2768
    },
    {
      "epoch": 1.1062724730323612,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0775,
      "step": 2769
    },
    {
      "epoch": 1.1066719936076708,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0852,
      "step": 2770
    },
    {
      "epoch": 1.1070715141829803,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 2771
    },
    {
      "epoch": 1.1074710347582901,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0743,
      "step": 2772
    },
    {
      "epoch": 1.1078705553335997,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0827,
      "step": 2773
    },
    {
      "epoch": 1.1082700759089092,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 2774
    },
    {
      "epoch": 1.108669596484219,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.076,
      "step": 2775
    },
    {
      "epoch": 1.1090691170595286,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0786,
      "step": 2776
    },
    {
      "epoch": 1.1094686376348382,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0005,
      "loss": 1.0798,
      "step": 2777
    },
    {
      "epoch": 1.1098681582101477,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 2778
    },
    {
      "epoch": 1.1102676787854575,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0005,
      "loss": 1.0765,
      "step": 2779
    },
    {
      "epoch": 1.110667199360767,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 2780
    },
    {
      "epoch": 1.1110667199360766,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0779,
      "step": 2781
    },
    {
      "epoch": 1.1114662405113864,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0798,
      "step": 2782
    },
    {
      "epoch": 1.111865761086696,
      "grad_norm": 0.171875,
      "learning_rate": 0.0005,
      "loss": 1.0843,
      "step": 2783
    },
    {
      "epoch": 1.1122652816620056,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 2784
    },
    {
      "epoch": 1.1126648022373151,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0005,
      "loss": 1.0737,
      "step": 2785
    },
    {
      "epoch": 1.113064322812625,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0856,
      "step": 2786
    },
    {
      "epoch": 1.1134638433879345,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0838,
      "step": 2787
    },
    {
      "epoch": 1.113863363963244,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0805,
      "step": 2788
    },
    {
      "epoch": 1.1142628845385538,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.0811,
      "step": 2789
    },
    {
      "epoch": 1.1146624051138634,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0791,
      "step": 2790
    },
    {
      "epoch": 1.115061925689173,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0005,
      "loss": 1.0877,
      "step": 2791
    },
    {
      "epoch": 1.1154614462644825,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0841,
      "step": 2792
    },
    {
      "epoch": 1.1158609668397923,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.0783,
      "step": 2793
    },
    {
      "epoch": 1.1162604874151019,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0774,
      "step": 2794
    },
    {
      "epoch": 1.1166600079904114,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0005,
      "loss": 1.0832,
      "step": 2795
    },
    {
      "epoch": 1.1170595285657212,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0693,
      "step": 2796
    },
    {
      "epoch": 1.1174590491410308,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 2797
    },
    {
      "epoch": 1.1178585697163403,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0722,
      "step": 2798
    },
    {
      "epoch": 1.1182580902916501,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0742,
      "step": 2799
    },
    {
      "epoch": 1.1186576108669597,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0863,
      "step": 2800
    },
    {
      "epoch": 1.1190571314422693,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0755,
      "step": 2801
    },
    {
      "epoch": 1.1194566520175788,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0838,
      "step": 2802
    },
    {
      "epoch": 1.1198561725928886,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 2803
    },
    {
      "epoch": 1.1202556931681982,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0865,
      "step": 2804
    },
    {
      "epoch": 1.1206552137435077,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0822,
      "step": 2805
    },
    {
      "epoch": 1.1210547343188173,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0827,
      "step": 2806
    },
    {
      "epoch": 1.121454254894127,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0777,
      "step": 2807
    },
    {
      "epoch": 1.1218537754694367,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 2808
    },
    {
      "epoch": 1.1222532960447462,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0811,
      "step": 2809
    },
    {
      "epoch": 1.122652816620056,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0749,
      "step": 2810
    },
    {
      "epoch": 1.1230523371953656,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 2811
    },
    {
      "epoch": 1.1234518577706751,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0848,
      "step": 2812
    },
    {
      "epoch": 1.123851378345985,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0803,
      "step": 2813
    },
    {
      "epoch": 1.1242508989212945,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 2814
    },
    {
      "epoch": 1.124650419496604,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0764,
      "step": 2815
    },
    {
      "epoch": 1.1250499400719136,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0781,
      "step": 2816
    },
    {
      "epoch": 1.1254494606472234,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 2817
    },
    {
      "epoch": 1.125848981222533,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0693,
      "step": 2818
    },
    {
      "epoch": 1.1262485017978425,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 2819
    },
    {
      "epoch": 1.1266480223731523,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0831,
      "step": 2820
    },
    {
      "epoch": 1.1270475429484619,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0814,
      "step": 2821
    },
    {
      "epoch": 1.1274470635237714,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0005,
      "loss": 1.0823,
      "step": 2822
    },
    {
      "epoch": 1.127846584099081,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0813,
      "step": 2823
    },
    {
      "epoch": 1.1282461046743908,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 2824
    },
    {
      "epoch": 1.1286456252497004,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 2825
    },
    {
      "epoch": 1.12904514582501,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0834,
      "step": 2826
    },
    {
      "epoch": 1.1294446664003197,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 2827
    },
    {
      "epoch": 1.1298441869756293,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0829,
      "step": 2828
    },
    {
      "epoch": 1.1302437075509388,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0759,
      "step": 2829
    },
    {
      "epoch": 1.1306432281262486,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0794,
      "step": 2830
    },
    {
      "epoch": 1.1310427487015582,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0835,
      "step": 2831
    },
    {
      "epoch": 1.1314422692768678,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0832,
      "step": 2832
    },
    {
      "epoch": 1.1318417898521773,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 2833
    },
    {
      "epoch": 1.132241310427487,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.077,
      "step": 2834
    },
    {
      "epoch": 1.1326408310027967,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 2835
    },
    {
      "epoch": 1.1330403515781062,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 2836
    },
    {
      "epoch": 1.1334398721534158,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0822,
      "step": 2837
    },
    {
      "epoch": 1.1338393927287256,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0847,
      "step": 2838
    },
    {
      "epoch": 1.1342389133040351,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0771,
      "step": 2839
    },
    {
      "epoch": 1.1346384338793447,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 2840
    },
    {
      "epoch": 1.1350379544546545,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 2841
    },
    {
      "epoch": 1.135437475029964,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.0781,
      "step": 2842
    },
    {
      "epoch": 1.1358369956052736,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0732,
      "step": 2843
    },
    {
      "epoch": 1.1362365161805834,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0849,
      "step": 2844
    },
    {
      "epoch": 1.136636036755893,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0815,
      "step": 2845
    },
    {
      "epoch": 1.1370355573312025,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0005,
      "loss": 1.0848,
      "step": 2846
    },
    {
      "epoch": 1.137435077906512,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0719,
      "step": 2847
    },
    {
      "epoch": 1.137834598481822,
      "grad_norm": 0.28125,
      "learning_rate": 0.0005,
      "loss": 1.0828,
      "step": 2848
    },
    {
      "epoch": 1.1382341190571315,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 2849
    },
    {
      "epoch": 1.138633639632441,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.0738,
      "step": 2850
    },
    {
      "epoch": 1.1390331602077506,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0005,
      "loss": 1.0882,
      "step": 2851
    },
    {
      "epoch": 1.1394326807830604,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0738,
      "step": 2852
    },
    {
      "epoch": 1.13983220135837,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0005,
      "loss": 1.0817,
      "step": 2853
    },
    {
      "epoch": 1.1402317219336795,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 2854
    },
    {
      "epoch": 1.1406312425089893,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0005,
      "loss": 1.0712,
      "step": 2855
    },
    {
      "epoch": 1.1410307630842988,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0005,
      "loss": 1.0807,
      "step": 2856
    },
    {
      "epoch": 1.1414302836596084,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005,
      "loss": 1.0864,
      "step": 2857
    },
    {
      "epoch": 1.1418298042349182,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0754,
      "step": 2858
    },
    {
      "epoch": 1.1422293248102278,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0005,
      "loss": 1.0772,
      "step": 2859
    },
    {
      "epoch": 1.1426288453855373,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0761,
      "step": 2860
    },
    {
      "epoch": 1.143028365960847,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0005,
      "loss": 1.0842,
      "step": 2861
    },
    {
      "epoch": 1.1434278865361567,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0005,
      "loss": 1.0793,
      "step": 2862
    },
    {
      "epoch": 1.1438274071114662,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0803,
      "step": 2863
    },
    {
      "epoch": 1.1442269276867758,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0793,
      "step": 2864
    },
    {
      "epoch": 1.1446264482620856,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 2865
    },
    {
      "epoch": 1.1450259688373952,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005,
      "loss": 1.0765,
      "step": 2866
    },
    {
      "epoch": 1.1454254894127047,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0795,
      "step": 2867
    },
    {
      "epoch": 1.1458250099880143,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 2868
    },
    {
      "epoch": 1.146224530563324,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0805,
      "step": 2869
    },
    {
      "epoch": 1.1466240511386336,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0005,
      "loss": 1.0754,
      "step": 2870
    },
    {
      "epoch": 1.1470235717139432,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0782,
      "step": 2871
    },
    {
      "epoch": 1.147423092289253,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 2872
    },
    {
      "epoch": 1.1478226128645626,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0764,
      "step": 2873
    },
    {
      "epoch": 1.1482221334398721,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0789,
      "step": 2874
    },
    {
      "epoch": 1.148621654015182,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0876,
      "step": 2875
    },
    {
      "epoch": 1.1490211745904915,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 2876
    },
    {
      "epoch": 1.149420695165801,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0787,
      "step": 2877
    },
    {
      "epoch": 1.1498202157411106,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0956,
      "step": 2878
    },
    {
      "epoch": 1.1502197363164204,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0772,
      "step": 2879
    },
    {
      "epoch": 1.15061925689173,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 2880
    },
    {
      "epoch": 1.1510187774670395,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 2881
    },
    {
      "epoch": 1.151418298042349,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0005,
      "loss": 1.0742,
      "step": 2882
    },
    {
      "epoch": 1.1518178186176589,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.08,
      "step": 2883
    },
    {
      "epoch": 1.1522173391929684,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.0794,
      "step": 2884
    },
    {
      "epoch": 1.152616859768278,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.083,
      "step": 2885
    },
    {
      "epoch": 1.1530163803435878,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.076,
      "step": 2886
    },
    {
      "epoch": 1.1534159009188973,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.086,
      "step": 2887
    },
    {
      "epoch": 1.153815421494207,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.085,
      "step": 2888
    },
    {
      "epoch": 1.1542149420695167,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0801,
      "step": 2889
    },
    {
      "epoch": 1.1546144626448263,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0762,
      "step": 2890
    },
    {
      "epoch": 1.1550139832201358,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 2891
    },
    {
      "epoch": 1.1554135037954454,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0817,
      "step": 2892
    },
    {
      "epoch": 1.1558130243707552,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.073,
      "step": 2893
    },
    {
      "epoch": 1.1562125449460647,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0856,
      "step": 2894
    },
    {
      "epoch": 1.1566120655213743,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0711,
      "step": 2895
    },
    {
      "epoch": 1.1570115860966839,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0796,
      "step": 2896
    },
    {
      "epoch": 1.1574111066719937,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0693,
      "step": 2897
    },
    {
      "epoch": 1.1578106272473032,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0745,
      "step": 2898
    },
    {
      "epoch": 1.1582101478226128,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 2899
    },
    {
      "epoch": 1.1586096683979226,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0809,
      "step": 2900
    },
    {
      "epoch": 1.1590091889732321,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0774,
      "step": 2901
    },
    {
      "epoch": 1.1594087095485417,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 2902
    },
    {
      "epoch": 1.1598082301238515,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 2903
    },
    {
      "epoch": 1.160207750699161,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0778,
      "step": 2904
    },
    {
      "epoch": 1.1606072712744706,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 2905
    },
    {
      "epoch": 1.1610067918497802,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 2906
    },
    {
      "epoch": 1.16140631242509,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 2907
    },
    {
      "epoch": 1.1618058330003995,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0809,
      "step": 2908
    },
    {
      "epoch": 1.162205353575709,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0863,
      "step": 2909
    },
    {
      "epoch": 1.1626048741510189,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 2910
    },
    {
      "epoch": 1.1630043947263284,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0735,
      "step": 2911
    },
    {
      "epoch": 1.163403915301638,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0715,
      "step": 2912
    },
    {
      "epoch": 1.1638034358769476,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 2913
    },
    {
      "epoch": 1.1642029564522574,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0789,
      "step": 2914
    },
    {
      "epoch": 1.164602477027567,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0719,
      "step": 2915
    },
    {
      "epoch": 1.1650019976028765,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0768,
      "step": 2916
    },
    {
      "epoch": 1.1654015181781863,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 2917
    },
    {
      "epoch": 1.1658010387534958,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.074,
      "step": 2918
    },
    {
      "epoch": 1.1662005593288054,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0914,
      "step": 2919
    },
    {
      "epoch": 1.1666000799041152,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0867,
      "step": 2920
    },
    {
      "epoch": 1.1669996004794247,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0796,
      "step": 2921
    },
    {
      "epoch": 1.1673991210547343,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 2922
    },
    {
      "epoch": 1.1677986416300439,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 2923
    },
    {
      "epoch": 1.1681981622053537,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0743,
      "step": 2924
    },
    {
      "epoch": 1.1685976827806632,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0737,
      "step": 2925
    },
    {
      "epoch": 1.1689972033559728,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0697,
      "step": 2926
    },
    {
      "epoch": 1.1693967239312824,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0795,
      "step": 2927
    },
    {
      "epoch": 1.1697962445065921,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0768,
      "step": 2928
    },
    {
      "epoch": 1.1701957650819017,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 2929
    },
    {
      "epoch": 1.1705952856572113,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0739,
      "step": 2930
    },
    {
      "epoch": 1.170994806232521,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0807,
      "step": 2931
    },
    {
      "epoch": 1.1713943268078306,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 2932
    },
    {
      "epoch": 1.1717938473831402,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 2933
    },
    {
      "epoch": 1.17219336795845,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 2934
    },
    {
      "epoch": 1.1725928885337595,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0753,
      "step": 2935
    },
    {
      "epoch": 1.172992409109069,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 2936
    },
    {
      "epoch": 1.1733919296843787,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0786,
      "step": 2937
    },
    {
      "epoch": 1.1737914502596885,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0906,
      "step": 2938
    },
    {
      "epoch": 1.174190970834998,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.076,
      "step": 2939
    },
    {
      "epoch": 1.1745904914103076,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0806,
      "step": 2940
    },
    {
      "epoch": 1.1749900119856171,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 2941
    },
    {
      "epoch": 1.175389532560927,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0803,
      "step": 2942
    },
    {
      "epoch": 1.1757890531362365,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0715,
      "step": 2943
    },
    {
      "epoch": 1.176188573711546,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0832,
      "step": 2944
    },
    {
      "epoch": 1.1765880942868558,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 2945
    },
    {
      "epoch": 1.1769876148621654,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 2946
    },
    {
      "epoch": 1.177387135437475,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 2947
    },
    {
      "epoch": 1.1777866560127848,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0804,
      "step": 2948
    },
    {
      "epoch": 1.1781861765880943,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0826,
      "step": 2949
    },
    {
      "epoch": 1.178585697163404,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0836,
      "step": 2950
    },
    {
      "epoch": 1.1789852177387135,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 2951
    },
    {
      "epoch": 1.1793847383140232,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 2952
    },
    {
      "epoch": 1.1797842588893328,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 2953
    },
    {
      "epoch": 1.1801837794646424,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0734,
      "step": 2954
    },
    {
      "epoch": 1.180583300039952,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0714,
      "step": 2955
    },
    {
      "epoch": 1.1809828206152617,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 2956
    },
    {
      "epoch": 1.1813823411905713,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.085,
      "step": 2957
    },
    {
      "epoch": 1.1817818617658808,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0757,
      "step": 2958
    },
    {
      "epoch": 1.1821813823411906,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0859,
      "step": 2959
    },
    {
      "epoch": 1.1825809029165002,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 2960
    },
    {
      "epoch": 1.1829804234918098,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.076,
      "step": 2961
    },
    {
      "epoch": 1.1833799440671195,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 2962
    },
    {
      "epoch": 1.1837794646424291,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 2963
    },
    {
      "epoch": 1.1841789852177387,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0005,
      "loss": 1.0898,
      "step": 2964
    },
    {
      "epoch": 1.1845785057930485,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 2965
    },
    {
      "epoch": 1.184978026368358,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.076,
      "step": 2966
    },
    {
      "epoch": 1.1853775469436676,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 2967
    },
    {
      "epoch": 1.1857770675189772,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0815,
      "step": 2968
    },
    {
      "epoch": 1.186176588094287,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0756,
      "step": 2969
    },
    {
      "epoch": 1.1865761086695965,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 2970
    },
    {
      "epoch": 1.186975629244906,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.078,
      "step": 2971
    },
    {
      "epoch": 1.1873751498202156,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 2972
    },
    {
      "epoch": 1.1877746703955254,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 2973
    },
    {
      "epoch": 1.188174190970835,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 2974
    },
    {
      "epoch": 1.1885737115461446,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0846,
      "step": 2975
    },
    {
      "epoch": 1.1889732321214543,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.087,
      "step": 2976
    },
    {
      "epoch": 1.189372752696764,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 2977
    },
    {
      "epoch": 1.1897722732720735,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 2978
    },
    {
      "epoch": 1.1901717938473833,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 2979
    },
    {
      "epoch": 1.1905713144226928,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 2980
    },
    {
      "epoch": 1.1909708349980024,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0761,
      "step": 2981
    },
    {
      "epoch": 1.191370355573312,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0756,
      "step": 2982
    },
    {
      "epoch": 1.1917698761486217,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0834,
      "step": 2983
    },
    {
      "epoch": 1.1921693967239313,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0865,
      "step": 2984
    },
    {
      "epoch": 1.1925689172992409,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 2985
    },
    {
      "epoch": 1.1929684378745504,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0731,
      "step": 2986
    },
    {
      "epoch": 1.1933679584498602,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 2987
    },
    {
      "epoch": 1.1937674790251698,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0786,
      "step": 2988
    },
    {
      "epoch": 1.1941669996004793,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 2989
    },
    {
      "epoch": 1.1945665201757891,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0005,
      "loss": 1.0767,
      "step": 2990
    },
    {
      "epoch": 1.1949660407510987,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 2991
    },
    {
      "epoch": 1.1953655613264083,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0815,
      "step": 2992
    },
    {
      "epoch": 1.195765081901718,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005,
      "loss": 1.0761,
      "step": 2993
    },
    {
      "epoch": 1.1961646024770276,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0005,
      "loss": 1.0736,
      "step": 2994
    },
    {
      "epoch": 1.1965641230523372,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 2995
    },
    {
      "epoch": 1.1969636436276467,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 2996
    },
    {
      "epoch": 1.1973631642029565,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0746,
      "step": 2997
    },
    {
      "epoch": 1.197762684778266,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005,
      "loss": 1.0858,
      "step": 2998
    },
    {
      "epoch": 1.1981622053535756,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0755,
      "step": 2999
    },
    {
      "epoch": 1.1985617259288852,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0005,
      "loss": 1.0816,
      "step": 3000
    },
    {
      "epoch": 1.198961246504195,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0005,
      "loss": 1.0763,
      "step": 3001
    },
    {
      "epoch": 1.1993607670795046,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 3002
    },
    {
      "epoch": 1.1997602876548141,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005,
      "loss": 1.0821,
      "step": 3003
    },
    {
      "epoch": 1.200159808230124,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0884,
      "step": 3004
    },
    {
      "epoch": 1.2005593288054335,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 3005
    },
    {
      "epoch": 1.200958849380743,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 3006
    },
    {
      "epoch": 1.2013583699560528,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 3007
    },
    {
      "epoch": 1.2017578905313624,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 3008
    },
    {
      "epoch": 1.202157411106672,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0736,
      "step": 3009
    },
    {
      "epoch": 1.2025569316819817,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0005,
      "loss": 1.0813,
      "step": 3010
    },
    {
      "epoch": 1.2029564522572913,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.085,
      "step": 3011
    },
    {
      "epoch": 1.2033559728326009,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0772,
      "step": 3012
    },
    {
      "epoch": 1.2037554934079104,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0005,
      "loss": 1.0817,
      "step": 3013
    },
    {
      "epoch": 1.2041550139832202,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 3014
    },
    {
      "epoch": 1.2045545345585298,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0005,
      "loss": 1.073,
      "step": 3015
    },
    {
      "epoch": 1.2049540551338394,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 3016
    },
    {
      "epoch": 1.205353575709149,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0005,
      "loss": 1.0853,
      "step": 3017
    },
    {
      "epoch": 1.2057530962844587,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0743,
      "step": 3018
    },
    {
      "epoch": 1.2061526168597683,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 3019
    },
    {
      "epoch": 1.2065521374350778,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 3020
    },
    {
      "epoch": 1.2069516580103876,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005,
      "loss": 1.0809,
      "step": 3021
    },
    {
      "epoch": 1.2073511785856972,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0679,
      "step": 3022
    },
    {
      "epoch": 1.2077506991610067,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.0764,
      "step": 3023
    },
    {
      "epoch": 1.2081502197363165,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0756,
      "step": 3024
    },
    {
      "epoch": 1.208549740311626,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0746,
      "step": 3025
    },
    {
      "epoch": 1.2089492608869357,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 3026
    },
    {
      "epoch": 1.2093487814622452,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 3027
    },
    {
      "epoch": 1.209748302037555,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.073,
      "step": 3028
    },
    {
      "epoch": 1.2101478226128646,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.088,
      "step": 3029
    },
    {
      "epoch": 1.2105473431881741,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0743,
      "step": 3030
    },
    {
      "epoch": 1.2109468637634837,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 3031
    },
    {
      "epoch": 1.2113463843387935,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0834,
      "step": 3032
    },
    {
      "epoch": 1.211745904914103,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 3033
    },
    {
      "epoch": 1.2121454254894126,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 3034
    },
    {
      "epoch": 1.2125449460647224,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.07,
      "step": 3035
    },
    {
      "epoch": 1.212944466640032,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0732,
      "step": 3036
    },
    {
      "epoch": 1.2133439872153415,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 3037
    },
    {
      "epoch": 1.2137435077906513,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0739,
      "step": 3038
    },
    {
      "epoch": 1.2141430283659609,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0744,
      "step": 3039
    },
    {
      "epoch": 1.2145425489412705,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.074,
      "step": 3040
    },
    {
      "epoch": 1.21494206951658,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0825,
      "step": 3041
    },
    {
      "epoch": 1.2153415900918898,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 3042
    },
    {
      "epoch": 1.2157411106671994,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0742,
      "step": 3043
    },
    {
      "epoch": 1.216140631242509,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0745,
      "step": 3044
    },
    {
      "epoch": 1.2165401518178185,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0786,
      "step": 3045
    },
    {
      "epoch": 1.2169396723931283,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0812,
      "step": 3046
    },
    {
      "epoch": 1.2173391929684378,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 3047
    },
    {
      "epoch": 1.2177387135437474,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0793,
      "step": 3048
    },
    {
      "epoch": 1.2181382341190572,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 3049
    },
    {
      "epoch": 1.2185377546943668,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0759,
      "step": 3050
    },
    {
      "epoch": 1.2189372752696763,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0944,
      "step": 3051
    },
    {
      "epoch": 1.219336795844986,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 3052
    },
    {
      "epoch": 1.2197363164202957,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 3053
    },
    {
      "epoch": 1.2201358369956052,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 3054
    },
    {
      "epoch": 1.220535357570915,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0812,
      "step": 3055
    },
    {
      "epoch": 1.2209348781462246,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0829,
      "step": 3056
    },
    {
      "epoch": 1.2213343987215342,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0754,
      "step": 3057
    },
    {
      "epoch": 1.2217339192968437,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 3058
    },
    {
      "epoch": 1.2221334398721535,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0779,
      "step": 3059
    },
    {
      "epoch": 1.222532960447463,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0756,
      "step": 3060
    },
    {
      "epoch": 1.2229324810227726,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 3061
    },
    {
      "epoch": 1.2233320015980822,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0815,
      "step": 3062
    },
    {
      "epoch": 1.223731522173392,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0775,
      "step": 3063
    },
    {
      "epoch": 1.2241310427487015,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 3064
    },
    {
      "epoch": 1.2245305633240111,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0735,
      "step": 3065
    },
    {
      "epoch": 1.224930083899321,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0715,
      "step": 3066
    },
    {
      "epoch": 1.2253296044746305,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0774,
      "step": 3067
    },
    {
      "epoch": 1.22572912504994,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 3068
    },
    {
      "epoch": 1.2261286456252498,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 3069
    },
    {
      "epoch": 1.2265281662005594,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0784,
      "step": 3070
    },
    {
      "epoch": 1.226927686775869,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0979,
      "step": 3071
    },
    {
      "epoch": 1.2273272073511785,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0719,
      "step": 3072
    },
    {
      "epoch": 1.2277267279264883,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0904,
      "step": 3073
    },
    {
      "epoch": 1.2281262485017979,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 3074
    },
    {
      "epoch": 1.2285257690771074,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 3075
    },
    {
      "epoch": 1.228925289652417,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0789,
      "step": 3076
    },
    {
      "epoch": 1.2293248102277268,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0743,
      "step": 3077
    },
    {
      "epoch": 1.2297243308030363,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 3078
    },
    {
      "epoch": 1.230123851378346,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0779,
      "step": 3079
    },
    {
      "epoch": 1.2305233719536557,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 3080
    },
    {
      "epoch": 1.2309228925289653,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 3081
    },
    {
      "epoch": 1.2313224131042748,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 3082
    },
    {
      "epoch": 1.2317219336795846,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 3083
    },
    {
      "epoch": 1.2321214542548942,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0712,
      "step": 3084
    },
    {
      "epoch": 1.2325209748302037,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0005,
      "loss": 1.0743,
      "step": 3085
    },
    {
      "epoch": 1.2329204954055133,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 3086
    },
    {
      "epoch": 1.233320015980823,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0679,
      "step": 3087
    },
    {
      "epoch": 1.2337195365561326,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 3088
    },
    {
      "epoch": 1.2341190571314422,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0781,
      "step": 3089
    },
    {
      "epoch": 1.2345185777067518,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 3090
    },
    {
      "epoch": 1.2349180982820616,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 3091
    },
    {
      "epoch": 1.2353176188573711,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0005,
      "loss": 1.0712,
      "step": 3092
    },
    {
      "epoch": 1.2357171394326807,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 3093
    },
    {
      "epoch": 1.2361166600079905,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 3094
    },
    {
      "epoch": 1.2365161805833,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0746,
      "step": 3095
    },
    {
      "epoch": 1.2369157011586096,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005,
      "loss": 1.0765,
      "step": 3096
    },
    {
      "epoch": 1.2373152217339194,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0792,
      "step": 3097
    },
    {
      "epoch": 1.237714742309229,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0005,
      "loss": 1.0746,
      "step": 3098
    },
    {
      "epoch": 1.2381142628845385,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005,
      "loss": 1.0719,
      "step": 3099
    },
    {
      "epoch": 1.2385137834598483,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0813,
      "step": 3100
    },
    {
      "epoch": 1.2389133040351579,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0005,
      "loss": 1.0793,
      "step": 3101
    },
    {
      "epoch": 1.2393128246104674,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 3102
    },
    {
      "epoch": 1.239712345185777,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0005,
      "loss": 1.0741,
      "step": 3103
    },
    {
      "epoch": 1.2401118657610868,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 3104
    },
    {
      "epoch": 1.2405113863363963,
      "grad_norm": 0.21875,
      "learning_rate": 0.0005,
      "loss": 1.0796,
      "step": 3105
    },
    {
      "epoch": 1.240910906911706,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0005,
      "loss": 1.0853,
      "step": 3106
    },
    {
      "epoch": 1.2413104274870155,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 3107
    },
    {
      "epoch": 1.2417099480623253,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005,
      "loss": 1.0821,
      "step": 3108
    },
    {
      "epoch": 1.2421094686376348,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0769,
      "step": 3109
    },
    {
      "epoch": 1.2425089892129444,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 3110
    },
    {
      "epoch": 1.2429085097882542,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 3111
    },
    {
      "epoch": 1.2433080303635637,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 3112
    },
    {
      "epoch": 1.2437075509388733,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.07,
      "step": 3113
    },
    {
      "epoch": 1.244107071514183,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 3114
    },
    {
      "epoch": 1.2445065920894927,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 3115
    },
    {
      "epoch": 1.2449061126648022,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0794,
      "step": 3116
    },
    {
      "epoch": 1.2453056332401118,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0815,
      "step": 3117
    },
    {
      "epoch": 1.2457051538154216,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0809,
      "step": 3118
    },
    {
      "epoch": 1.2461046743907311,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0802,
      "step": 3119
    },
    {
      "epoch": 1.2465041949660407,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0712,
      "step": 3120
    },
    {
      "epoch": 1.2469037155413503,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0732,
      "step": 3121
    },
    {
      "epoch": 1.24730323611666,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 3122
    },
    {
      "epoch": 1.2477027566919696,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0739,
      "step": 3123
    },
    {
      "epoch": 1.2481022772672792,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0742,
      "step": 3124
    },
    {
      "epoch": 1.248501797842589,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0839,
      "step": 3125
    },
    {
      "epoch": 1.2489013184178985,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0908,
      "step": 3126
    },
    {
      "epoch": 1.249300838993208,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 3127
    },
    {
      "epoch": 1.2497003595685179,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0827,
      "step": 3128
    },
    {
      "epoch": 1.2500998801438274,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 3129
    },
    {
      "epoch": 1.250499400719137,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.086,
      "step": 3130
    },
    {
      "epoch": 1.2508989212944468,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0761,
      "step": 3131
    },
    {
      "epoch": 1.2512984418697564,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0804,
      "step": 3132
    },
    {
      "epoch": 1.251697962445066,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0842,
      "step": 3133
    },
    {
      "epoch": 1.2520974830203755,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 3134
    },
    {
      "epoch": 1.252497003595685,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0722,
      "step": 3135
    },
    {
      "epoch": 1.2528965241709948,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 3136
    },
    {
      "epoch": 1.2532960447463044,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0651,
      "step": 3137
    },
    {
      "epoch": 1.253695565321614,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0712,
      "step": 3138
    },
    {
      "epoch": 1.2540950858969238,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0768,
      "step": 3139
    },
    {
      "epoch": 1.2544946064722333,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 3140
    },
    {
      "epoch": 1.2548941270475429,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0704,
      "step": 3141
    },
    {
      "epoch": 1.2552936476228527,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 3142
    },
    {
      "epoch": 1.2556931681981622,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0824,
      "step": 3143
    },
    {
      "epoch": 1.2560926887734718,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.079,
      "step": 3144
    },
    {
      "epoch": 1.2564922093487816,
      "grad_norm": 0.203125,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 3145
    },
    {
      "epoch": 1.2568917299240912,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.085,
      "step": 3146
    },
    {
      "epoch": 1.2572912504994007,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0005,
      "loss": 1.0798,
      "step": 3147
    },
    {
      "epoch": 1.2576907710747103,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 3148
    },
    {
      "epoch": 1.2580902916500198,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.078,
      "step": 3149
    },
    {
      "epoch": 1.2584898122253296,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0772,
      "step": 3150
    },
    {
      "epoch": 1.2588893328006392,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 3151
    },
    {
      "epoch": 1.2592888533759488,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0005,
      "loss": 1.0786,
      "step": 3152
    },
    {
      "epoch": 1.2596883739512585,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 3153
    },
    {
      "epoch": 1.260087894526568,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0005,
      "loss": 1.0657,
      "step": 3154
    },
    {
      "epoch": 1.2604874151018777,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 3155
    },
    {
      "epoch": 1.2608869356771875,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 3156
    },
    {
      "epoch": 1.261286456252497,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0819,
      "step": 3157
    },
    {
      "epoch": 1.2616859768278066,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0732,
      "step": 3158
    },
    {
      "epoch": 1.2620854974031164,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 3159
    },
    {
      "epoch": 1.262485017978426,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0743,
      "step": 3160
    },
    {
      "epoch": 1.2628845385537355,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 3161
    },
    {
      "epoch": 1.263284059129045,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 3162
    },
    {
      "epoch": 1.2636835797043549,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0794,
      "step": 3163
    },
    {
      "epoch": 1.2640831002796644,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.0696,
      "step": 3164
    },
    {
      "epoch": 1.264482620854974,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0823,
      "step": 3165
    },
    {
      "epoch": 1.2648821414302835,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0809,
      "step": 3166
    },
    {
      "epoch": 1.2652816620055933,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0744,
      "step": 3167
    },
    {
      "epoch": 1.265681182580903,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.073,
      "step": 3168
    },
    {
      "epoch": 1.2660807031562125,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.07,
      "step": 3169
    },
    {
      "epoch": 1.2664802237315222,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 3170
    },
    {
      "epoch": 1.2668797443068318,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0732,
      "step": 3171
    },
    {
      "epoch": 1.2672792648821414,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0769,
      "step": 3172
    },
    {
      "epoch": 1.2676787854574512,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0863,
      "step": 3173
    },
    {
      "epoch": 1.2680783060327607,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 3174
    },
    {
      "epoch": 1.2684778266080703,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 3175
    },
    {
      "epoch": 1.26887734718338,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0741,
      "step": 3176
    },
    {
      "epoch": 1.2692768677586896,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0735,
      "step": 3177
    },
    {
      "epoch": 1.2696763883339992,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0774,
      "step": 3178
    },
    {
      "epoch": 1.2700759089093088,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0728,
      "step": 3179
    },
    {
      "epoch": 1.2704754294846183,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0721,
      "step": 3180
    },
    {
      "epoch": 1.2708749500599281,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.075,
      "step": 3181
    },
    {
      "epoch": 1.2712744706352377,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0775,
      "step": 3182
    },
    {
      "epoch": 1.2716739912105472,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0777,
      "step": 3183
    },
    {
      "epoch": 1.272073511785857,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0799,
      "step": 3184
    },
    {
      "epoch": 1.2724730323611666,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0926,
      "step": 3185
    },
    {
      "epoch": 1.2728725529364762,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.078,
      "step": 3186
    },
    {
      "epoch": 1.273272073511786,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 3187
    },
    {
      "epoch": 1.2736715940870955,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0808,
      "step": 3188
    },
    {
      "epoch": 1.274071114662405,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0767,
      "step": 3189
    },
    {
      "epoch": 1.2744706352377149,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 3190
    },
    {
      "epoch": 1.2748701558130244,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 3191
    },
    {
      "epoch": 1.275269676388334,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 3192
    },
    {
      "epoch": 1.2756691969636436,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 3193
    },
    {
      "epoch": 1.2760687175389531,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 3194
    },
    {
      "epoch": 1.276468238114263,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 3195
    },
    {
      "epoch": 1.2768677586895725,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0717,
      "step": 3196
    },
    {
      "epoch": 1.277267279264882,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0885,
      "step": 3197
    },
    {
      "epoch": 1.2776667998401918,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 3198
    },
    {
      "epoch": 1.2780663204155014,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 3199
    },
    {
      "epoch": 1.278465840990811,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 3200
    },
    {
      "epoch": 1.2788653615661207,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 3201
    },
    {
      "epoch": 1.2792648821414303,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 3202
    },
    {
      "epoch": 1.2796644027167399,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0786,
      "step": 3203
    },
    {
      "epoch": 1.2800639232920497,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0842,
      "step": 3204
    },
    {
      "epoch": 1.2804634438673592,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 3205
    },
    {
      "epoch": 1.2808629644426688,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0807,
      "step": 3206
    },
    {
      "epoch": 1.2812624850179783,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0697,
      "step": 3207
    },
    {
      "epoch": 1.2816620055932881,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0784,
      "step": 3208
    },
    {
      "epoch": 1.2820615261685977,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0789,
      "step": 3209
    },
    {
      "epoch": 1.2824610467439073,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0824,
      "step": 3210
    },
    {
      "epoch": 1.2828605673192168,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 3211
    },
    {
      "epoch": 1.2832600878945266,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0776,
      "step": 3212
    },
    {
      "epoch": 1.2836596084698362,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0716,
      "step": 3213
    },
    {
      "epoch": 1.2840591290451457,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0781,
      "step": 3214
    },
    {
      "epoch": 1.2844586496204555,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 3215
    },
    {
      "epoch": 1.284858170195765,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0736,
      "step": 3216
    },
    {
      "epoch": 1.2852576907710747,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0777,
      "step": 3217
    },
    {
      "epoch": 1.2856572113463844,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0715,
      "step": 3218
    },
    {
      "epoch": 1.286056731921694,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 3219
    },
    {
      "epoch": 1.2864562524970036,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0804,
      "step": 3220
    },
    {
      "epoch": 1.2868557730723134,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 3221
    },
    {
      "epoch": 1.287255293647623,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 3222
    },
    {
      "epoch": 1.2876548142229325,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 3223
    },
    {
      "epoch": 1.288054334798242,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 3224
    },
    {
      "epoch": 1.2884538553735516,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 3225
    },
    {
      "epoch": 1.2888533759488614,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0791,
      "step": 3226
    },
    {
      "epoch": 1.289252896524171,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 3227
    },
    {
      "epoch": 1.2896524170994805,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 3228
    },
    {
      "epoch": 1.2900519376747903,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0732,
      "step": 3229
    },
    {
      "epoch": 1.2904514582500999,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.073,
      "step": 3230
    },
    {
      "epoch": 1.2908509788254094,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0687,
      "step": 3231
    },
    {
      "epoch": 1.2912504994007192,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.078,
      "step": 3232
    },
    {
      "epoch": 1.2916500199760288,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0796,
      "step": 3233
    },
    {
      "epoch": 1.2920495405513384,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 3234
    },
    {
      "epoch": 1.2924490611266481,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 3235
    },
    {
      "epoch": 1.2928485817019577,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0779,
      "step": 3236
    },
    {
      "epoch": 1.2932481022772673,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0745,
      "step": 3237
    },
    {
      "epoch": 1.2936476228525768,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0725,
      "step": 3238
    },
    {
      "epoch": 1.2940471434278864,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0867,
      "step": 3239
    },
    {
      "epoch": 1.2944466640031962,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0005,
      "loss": 1.0761,
      "step": 3240
    },
    {
      "epoch": 1.2948461845785058,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.0823,
      "step": 3241
    },
    {
      "epoch": 1.2952457051538153,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0763,
      "step": 3242
    },
    {
      "epoch": 1.295645225729125,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 3243
    },
    {
      "epoch": 1.2960447463044347,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.084,
      "step": 3244
    },
    {
      "epoch": 1.2964442668797442,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0813,
      "step": 3245
    },
    {
      "epoch": 1.296843787455054,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0756,
      "step": 3246
    },
    {
      "epoch": 1.2972433080303636,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0777,
      "step": 3247
    },
    {
      "epoch": 1.2976428286056731,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 3248
    },
    {
      "epoch": 1.298042349180983,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0737,
      "step": 3249
    },
    {
      "epoch": 1.2984418697562925,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.079,
      "step": 3250
    },
    {
      "epoch": 1.298841390331602,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 3251
    },
    {
      "epoch": 1.2992409109069116,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 3252
    },
    {
      "epoch": 1.2996404314822214,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0755,
      "step": 3253
    },
    {
      "epoch": 1.300039952057531,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 3254
    },
    {
      "epoch": 1.3004394726328405,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0779,
      "step": 3255
    },
    {
      "epoch": 1.30083899320815,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 3256
    },
    {
      "epoch": 1.30123851378346,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 3257
    },
    {
      "epoch": 1.3016380343587695,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.082,
      "step": 3258
    },
    {
      "epoch": 1.302037554934079,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0802,
      "step": 3259
    },
    {
      "epoch": 1.3024370755093888,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 3260
    },
    {
      "epoch": 1.3028365960846984,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0753,
      "step": 3261
    },
    {
      "epoch": 1.303236116660008,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 3262
    },
    {
      "epoch": 1.3036356372353177,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 3263
    },
    {
      "epoch": 1.3040351578106273,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 3264
    },
    {
      "epoch": 1.3044346783859369,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0801,
      "step": 3265
    },
    {
      "epoch": 1.3048341989612466,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0798,
      "step": 3266
    },
    {
      "epoch": 1.3052337195365562,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 3267
    },
    {
      "epoch": 1.3056332401118658,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0754,
      "step": 3268
    },
    {
      "epoch": 1.3060327606871753,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 3269
    },
    {
      "epoch": 1.306432281262485,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0742,
      "step": 3270
    },
    {
      "epoch": 1.3068318018377947,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0598,
      "step": 3271
    },
    {
      "epoch": 1.3072313224131042,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0857,
      "step": 3272
    },
    {
      "epoch": 1.3076308429884138,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0783,
      "step": 3273
    },
    {
      "epoch": 1.3080303635637236,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 3274
    },
    {
      "epoch": 1.3084298841390332,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0775,
      "step": 3275
    },
    {
      "epoch": 1.3088294047143427,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 3276
    },
    {
      "epoch": 1.3092289252896525,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 3277
    },
    {
      "epoch": 1.309628445864962,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 3278
    },
    {
      "epoch": 1.3100279664402716,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0742,
      "step": 3279
    },
    {
      "epoch": 1.3104274870155814,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0806,
      "step": 3280
    },
    {
      "epoch": 1.310827007590891,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0728,
      "step": 3281
    },
    {
      "epoch": 1.3112265281662006,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0741,
      "step": 3282
    },
    {
      "epoch": 1.3116260487415101,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 3283
    },
    {
      "epoch": 1.3120255693168197,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0857,
      "step": 3284
    },
    {
      "epoch": 1.3124250898921295,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 3285
    },
    {
      "epoch": 1.312824610467439,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 3286
    },
    {
      "epoch": 1.3132241310427486,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0739,
      "step": 3287
    },
    {
      "epoch": 1.3136236516180584,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 3288
    },
    {
      "epoch": 1.314023172193368,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 3289
    },
    {
      "epoch": 1.3144226927686775,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 3290
    },
    {
      "epoch": 1.3148222133439873,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.073,
      "step": 3291
    },
    {
      "epoch": 1.3152217339192969,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0782,
      "step": 3292
    },
    {
      "epoch": 1.3156212544946064,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0734,
      "step": 3293
    },
    {
      "epoch": 1.3160207750699162,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0831,
      "step": 3294
    },
    {
      "epoch": 1.3164202956452258,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0738,
      "step": 3295
    },
    {
      "epoch": 1.3168198162205353,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0775,
      "step": 3296
    },
    {
      "epoch": 1.317219336795845,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 3297
    },
    {
      "epoch": 1.3176188573711547,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0735,
      "step": 3298
    },
    {
      "epoch": 1.3180183779464643,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 3299
    },
    {
      "epoch": 1.3184178985217738,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0722,
      "step": 3300
    },
    {
      "epoch": 1.3188174190970834,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 3301
    },
    {
      "epoch": 1.3192169396723932,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 3302
    },
    {
      "epoch": 1.3196164602477027,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0816,
      "step": 3303
    },
    {
      "epoch": 1.3200159808230123,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0752,
      "step": 3304
    },
    {
      "epoch": 1.320415501398322,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0731,
      "step": 3305
    },
    {
      "epoch": 1.3208150219736317,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0765,
      "step": 3306
    },
    {
      "epoch": 1.3212145425489412,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0727,
      "step": 3307
    },
    {
      "epoch": 1.321614063124251,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 3308
    },
    {
      "epoch": 1.3220135836995606,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0775,
      "step": 3309
    },
    {
      "epoch": 1.3224131042748701,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0818,
      "step": 3310
    },
    {
      "epoch": 1.32281262485018,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 3311
    },
    {
      "epoch": 1.3232121454254895,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 3312
    },
    {
      "epoch": 1.323611666000799,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 3313
    },
    {
      "epoch": 1.3240111865761086,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 3314
    },
    {
      "epoch": 1.3244107071514182,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0722,
      "step": 3315
    },
    {
      "epoch": 1.324810227726728,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 3316
    },
    {
      "epoch": 1.3252097483020375,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 3317
    },
    {
      "epoch": 1.325609268877347,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 3318
    },
    {
      "epoch": 1.3260087894526569,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0719,
      "step": 3319
    },
    {
      "epoch": 1.3264083100279664,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 3320
    },
    {
      "epoch": 1.326807830603276,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 3321
    },
    {
      "epoch": 1.3272073511785858,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0717,
      "step": 3322
    },
    {
      "epoch": 1.3276068717538954,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0657,
      "step": 3323
    },
    {
      "epoch": 1.328006392329205,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 3324
    },
    {
      "epoch": 1.3284059129045147,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0805,
      "step": 3325
    },
    {
      "epoch": 1.3288054334798243,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0711,
      "step": 3326
    },
    {
      "epoch": 1.3292049540551338,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0744,
      "step": 3327
    },
    {
      "epoch": 1.3296044746304434,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0788,
      "step": 3328
    },
    {
      "epoch": 1.330003995205753,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0742,
      "step": 3329
    },
    {
      "epoch": 1.3304035157810628,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 3330
    },
    {
      "epoch": 1.3308030363563723,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 3331
    },
    {
      "epoch": 1.3312025569316819,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 3332
    },
    {
      "epoch": 1.3316020775069917,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0878,
      "step": 3333
    },
    {
      "epoch": 1.3320015980823012,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 3334
    },
    {
      "epoch": 1.3324011186576108,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0871,
      "step": 3335
    },
    {
      "epoch": 1.3328006392329206,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 3336
    },
    {
      "epoch": 1.3332001598082301,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0756,
      "step": 3337
    },
    {
      "epoch": 1.3335996803835397,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.074,
      "step": 3338
    },
    {
      "epoch": 1.3339992009588495,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0801,
      "step": 3339
    },
    {
      "epoch": 1.334398721534159,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 3340
    },
    {
      "epoch": 1.3347982421094686,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0749,
      "step": 3341
    },
    {
      "epoch": 1.3351977626847782,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0815,
      "step": 3342
    },
    {
      "epoch": 1.335597283260088,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0806,
      "step": 3343
    },
    {
      "epoch": 1.3359968038353975,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 3344
    },
    {
      "epoch": 1.336396324410707,
      "grad_norm": 0.265625,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 3345
    },
    {
      "epoch": 1.3367958449860167,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 3346
    },
    {
      "epoch": 1.3371953655613265,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 3347
    },
    {
      "epoch": 1.337594886136636,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 3348
    },
    {
      "epoch": 1.3379944067119456,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 3349
    },
    {
      "epoch": 1.3383939272872554,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 3350
    },
    {
      "epoch": 1.338793447862565,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0739,
      "step": 3351
    },
    {
      "epoch": 1.3391929684378745,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 3352
    },
    {
      "epoch": 1.3395924890131843,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 3353
    },
    {
      "epoch": 1.3399920095884938,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 3354
    },
    {
      "epoch": 1.3403915301638034,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0735,
      "step": 3355
    },
    {
      "epoch": 1.3407910507391132,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.0736,
      "step": 3356
    },
    {
      "epoch": 1.3411905713144228,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0005,
      "loss": 1.0762,
      "step": 3357
    },
    {
      "epoch": 1.3415900918897323,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 3358
    },
    {
      "epoch": 1.341989612465042,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 3359
    },
    {
      "epoch": 1.3423891330403515,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 3360
    },
    {
      "epoch": 1.3427886536156612,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 3361
    },
    {
      "epoch": 1.3431881741909708,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 3362
    },
    {
      "epoch": 1.3435876947662804,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0823,
      "step": 3363
    },
    {
      "epoch": 1.3439872153415902,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 3364
    },
    {
      "epoch": 1.3443867359168997,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 3365
    },
    {
      "epoch": 1.3447862564922093,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0728,
      "step": 3366
    },
    {
      "epoch": 1.345185777067519,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 3367
    },
    {
      "epoch": 1.3455852976428286,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0735,
      "step": 3368
    },
    {
      "epoch": 1.3459848182181382,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0813,
      "step": 3369
    },
    {
      "epoch": 1.346384338793448,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.069,
      "step": 3370
    },
    {
      "epoch": 1.3467838593687576,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0716,
      "step": 3371
    },
    {
      "epoch": 1.3471833799440671,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 3372
    },
    {
      "epoch": 1.3475829005193767,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0741,
      "step": 3373
    },
    {
      "epoch": 1.3479824210946862,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0817,
      "step": 3374
    },
    {
      "epoch": 1.348381941669996,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0005,
      "loss": 1.0828,
      "step": 3375
    },
    {
      "epoch": 1.3487814622453056,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 3376
    },
    {
      "epoch": 1.3491809828206152,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 3377
    },
    {
      "epoch": 1.349580503395925,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0756,
      "step": 3378
    },
    {
      "epoch": 1.3499800239712345,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 3379
    },
    {
      "epoch": 1.350379544546544,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 3380
    },
    {
      "epoch": 1.3507790651218539,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0843,
      "step": 3381
    },
    {
      "epoch": 1.3511785856971634,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 3382
    },
    {
      "epoch": 1.351578106272473,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 3383
    },
    {
      "epoch": 1.3519776268477828,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 3384
    },
    {
      "epoch": 1.3523771474230923,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 3385
    },
    {
      "epoch": 1.352776667998402,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0696,
      "step": 3386
    },
    {
      "epoch": 1.3531761885737115,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0779,
      "step": 3387
    },
    {
      "epoch": 1.3535757091490213,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0749,
      "step": 3388
    },
    {
      "epoch": 1.3539752297243308,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 3389
    },
    {
      "epoch": 1.3543747502996404,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0775,
      "step": 3390
    },
    {
      "epoch": 1.35477427087495,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 3391
    },
    {
      "epoch": 1.3551737914502597,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0866,
      "step": 3392
    },
    {
      "epoch": 1.3555733120255693,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 3393
    },
    {
      "epoch": 1.3559728326008789,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 3394
    },
    {
      "epoch": 1.3563723531761886,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 3395
    },
    {
      "epoch": 1.3567718737514982,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 3396
    },
    {
      "epoch": 1.3571713943268078,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 3397
    },
    {
      "epoch": 1.3575709149021176,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0767,
      "step": 3398
    },
    {
      "epoch": 1.3579704354774271,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0005,
      "loss": 1.0788,
      "step": 3399
    },
    {
      "epoch": 1.3583699560527367,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 3400
    },
    {
      "epoch": 1.3587694766280465,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 3401
    },
    {
      "epoch": 1.359168997203356,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0842,
      "step": 3402
    },
    {
      "epoch": 1.3595685177786656,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.0745,
      "step": 3403
    },
    {
      "epoch": 1.3599680383539752,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0802,
      "step": 3404
    },
    {
      "epoch": 1.3603675589292847,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0005,
      "loss": 1.0788,
      "step": 3405
    },
    {
      "epoch": 1.3607670795045945,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.0832,
      "step": 3406
    },
    {
      "epoch": 1.361166600079904,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0672,
      "step": 3407
    },
    {
      "epoch": 1.3615661206552137,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0005,
      "loss": 1.081,
      "step": 3408
    },
    {
      "epoch": 1.3619656412305234,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0693,
      "step": 3409
    },
    {
      "epoch": 1.362365161805833,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005,
      "loss": 1.0755,
      "step": 3410
    },
    {
      "epoch": 1.3627646823811426,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0781,
      "step": 3411
    },
    {
      "epoch": 1.3631642029564524,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0005,
      "loss": 1.0826,
      "step": 3412
    },
    {
      "epoch": 1.363563723531762,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 3413
    },
    {
      "epoch": 1.3639632441070715,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.073,
      "step": 3414
    },
    {
      "epoch": 1.3643627646823813,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0794,
      "step": 3415
    },
    {
      "epoch": 1.3647622852576908,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 3416
    },
    {
      "epoch": 1.3651618058330004,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0744,
      "step": 3417
    },
    {
      "epoch": 1.36556132640831,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 3418
    },
    {
      "epoch": 1.3659608469836195,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 3419
    },
    {
      "epoch": 1.3663603675589293,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0804,
      "step": 3420
    },
    {
      "epoch": 1.3667598881342389,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0898,
      "step": 3421
    },
    {
      "epoch": 1.3671594087095484,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0821,
      "step": 3422
    },
    {
      "epoch": 1.3675589292848582,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 3423
    },
    {
      "epoch": 1.3679584498601678,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 3424
    },
    {
      "epoch": 1.3683579704354774,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 3425
    },
    {
      "epoch": 1.3687574910107871,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0005,
      "loss": 1.0804,
      "step": 3426
    },
    {
      "epoch": 1.3691570115860967,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 3427
    },
    {
      "epoch": 1.3695565321614063,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0755,
      "step": 3428
    },
    {
      "epoch": 1.369956052736716,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 3429
    },
    {
      "epoch": 1.3703555733120256,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0783,
      "step": 3430
    },
    {
      "epoch": 1.3707550938873352,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 3431
    },
    {
      "epoch": 1.3711546144626447,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 3432
    },
    {
      "epoch": 1.3715541350379545,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 3433
    },
    {
      "epoch": 1.371953655613264,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0904,
      "step": 3434
    },
    {
      "epoch": 1.3723531761885737,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 3435
    },
    {
      "epoch": 1.3727526967638832,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 3436
    },
    {
      "epoch": 1.373152217339193,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 3437
    },
    {
      "epoch": 1.3735517379145026,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 3438
    },
    {
      "epoch": 1.3739512584898121,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0839,
      "step": 3439
    },
    {
      "epoch": 1.374350779065122,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0795,
      "step": 3440
    },
    {
      "epoch": 1.3747502996404315,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0788,
      "step": 3441
    },
    {
      "epoch": 1.375149820215741,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0704,
      "step": 3442
    },
    {
      "epoch": 1.3755493407910508,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.073,
      "step": 3443
    },
    {
      "epoch": 1.3759488613663604,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 3444
    },
    {
      "epoch": 1.37634838194167,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 3445
    },
    {
      "epoch": 1.3767479025169798,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0754,
      "step": 3446
    },
    {
      "epoch": 1.3771474230922893,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0753,
      "step": 3447
    },
    {
      "epoch": 1.3775469436675989,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.078,
      "step": 3448
    },
    {
      "epoch": 1.3779464642429085,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 3449
    },
    {
      "epoch": 1.378345984818218,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 3450
    },
    {
      "epoch": 1.3787455053935278,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 3451
    },
    {
      "epoch": 1.3791450259688374,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0781,
      "step": 3452
    },
    {
      "epoch": 1.379544546544147,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0805,
      "step": 3453
    },
    {
      "epoch": 1.3799440671194567,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0752,
      "step": 3454
    },
    {
      "epoch": 1.3803435876947663,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0783,
      "step": 3455
    },
    {
      "epoch": 1.3807431082700758,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0772,
      "step": 3456
    },
    {
      "epoch": 1.3811426288453856,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0727,
      "step": 3457
    },
    {
      "epoch": 1.3815421494206952,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0749,
      "step": 3458
    },
    {
      "epoch": 1.3819416699960048,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.07,
      "step": 3459
    },
    {
      "epoch": 1.3823411905713145,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 3460
    },
    {
      "epoch": 1.3827407111466241,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.0665,
      "step": 3461
    },
    {
      "epoch": 1.3831402317219337,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0736,
      "step": 3462
    },
    {
      "epoch": 1.3835397522972432,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0005,
      "loss": 1.0728,
      "step": 3463
    },
    {
      "epoch": 1.3839392728725528,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 3464
    },
    {
      "epoch": 1.3843387934478626,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 3465
    },
    {
      "epoch": 1.3847383140231722,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 3466
    },
    {
      "epoch": 1.3851378345984817,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 3467
    },
    {
      "epoch": 1.3855373551737915,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 3468
    },
    {
      "epoch": 1.385936875749101,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 3469
    },
    {
      "epoch": 1.3863363963244106,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0735,
      "step": 3470
    },
    {
      "epoch": 1.3867359168997204,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0005,
      "loss": 1.0777,
      "step": 3471
    },
    {
      "epoch": 1.38713543747503,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 3472
    },
    {
      "epoch": 1.3875349580503396,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 3473
    },
    {
      "epoch": 1.3879344786256493,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 3474
    },
    {
      "epoch": 1.388333999200959,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 3475
    },
    {
      "epoch": 1.3887335197762685,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 3476
    },
    {
      "epoch": 1.389133040351578,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 3477
    },
    {
      "epoch": 1.3895325609268878,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0908,
      "step": 3478
    },
    {
      "epoch": 1.3899320815021974,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 3479
    },
    {
      "epoch": 1.390331602077507,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0715,
      "step": 3480
    },
    {
      "epoch": 1.3907311226528165,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0767,
      "step": 3481
    },
    {
      "epoch": 1.3911306432281263,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0787,
      "step": 3482
    },
    {
      "epoch": 1.3915301638034359,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 3483
    },
    {
      "epoch": 1.3919296843787454,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 3484
    },
    {
      "epoch": 1.3923292049540552,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0796,
      "step": 3485
    },
    {
      "epoch": 1.3927287255293648,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0743,
      "step": 3486
    },
    {
      "epoch": 1.3931282461046743,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0776,
      "step": 3487
    },
    {
      "epoch": 1.3935277666799841,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0824,
      "step": 3488
    },
    {
      "epoch": 1.3939272872552937,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0749,
      "step": 3489
    },
    {
      "epoch": 1.3943268078306033,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0736,
      "step": 3490
    },
    {
      "epoch": 1.394726328405913,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 3491
    },
    {
      "epoch": 1.3951258489812226,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0727,
      "step": 3492
    },
    {
      "epoch": 1.3955253695565322,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 3493
    },
    {
      "epoch": 1.3959248901318417,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0752,
      "step": 3494
    },
    {
      "epoch": 1.3963244107071513,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0795,
      "step": 3495
    },
    {
      "epoch": 1.396723931282461,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 3496
    },
    {
      "epoch": 1.3971234518577706,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 3497
    },
    {
      "epoch": 1.3975229724330802,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0782,
      "step": 3498
    },
    {
      "epoch": 1.39792249300839,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 3499
    },
    {
      "epoch": 1.3983220135836996,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 3500
    },
    {
      "epoch": 1.3987215341590091,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.083,
      "step": 3501
    },
    {
      "epoch": 1.399121054734319,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 3502
    },
    {
      "epoch": 1.3995205753096285,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 3503
    },
    {
      "epoch": 1.399920095884938,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 3504
    },
    {
      "epoch": 1.4003196164602478,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0801,
      "step": 3505
    },
    {
      "epoch": 1.4007191370355574,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0719,
      "step": 3506
    },
    {
      "epoch": 1.401118657610867,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 3507
    },
    {
      "epoch": 1.4015181781861765,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 3508
    },
    {
      "epoch": 1.401917698761486,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 3509
    },
    {
      "epoch": 1.4023172193367959,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 3510
    },
    {
      "epoch": 1.4027167399121054,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0771,
      "step": 3511
    },
    {
      "epoch": 1.403116260487415,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0716,
      "step": 3512
    },
    {
      "epoch": 1.4035157810627248,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 3513
    },
    {
      "epoch": 1.4039153016380344,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 3514
    },
    {
      "epoch": 1.404314822213344,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 3515
    },
    {
      "epoch": 1.4047143427886537,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0732,
      "step": 3516
    },
    {
      "epoch": 1.4051138633639633,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 3517
    },
    {
      "epoch": 1.4055133839392728,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 3518
    },
    {
      "epoch": 1.4059129045145826,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 3519
    },
    {
      "epoch": 1.4063124250898922,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0757,
      "step": 3520
    },
    {
      "epoch": 1.4067119456652017,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 3521
    },
    {
      "epoch": 1.4071114662405113,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0739,
      "step": 3522
    },
    {
      "epoch": 1.407510986815821,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 3523
    },
    {
      "epoch": 1.4079105073911307,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 3524
    },
    {
      "epoch": 1.4083100279664402,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 3525
    },
    {
      "epoch": 1.4087095485417498,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.079,
      "step": 3526
    },
    {
      "epoch": 1.4091090691170596,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 3527
    },
    {
      "epoch": 1.4095085896923691,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0768,
      "step": 3528
    },
    {
      "epoch": 1.4099081102676787,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0854,
      "step": 3529
    },
    {
      "epoch": 1.4103076308429885,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 3530
    },
    {
      "epoch": 1.410707151418298,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0697,
      "step": 3531
    },
    {
      "epoch": 1.4111066719936076,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 3532
    },
    {
      "epoch": 1.4115061925689174,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 3533
    },
    {
      "epoch": 1.411905713144227,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0749,
      "step": 3534
    },
    {
      "epoch": 1.4123052337195365,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0729,
      "step": 3535
    },
    {
      "epoch": 1.4127047542948463,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 3536
    },
    {
      "epoch": 1.4131042748701559,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0823,
      "step": 3537
    },
    {
      "epoch": 1.4135037954454654,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0793,
      "step": 3538
    },
    {
      "epoch": 1.413903316020775,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 3539
    },
    {
      "epoch": 1.4143028365960846,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 3540
    },
    {
      "epoch": 1.4147023571713944,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 3541
    },
    {
      "epoch": 1.415101877746704,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 3542
    },
    {
      "epoch": 1.4155013983220135,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 3543
    },
    {
      "epoch": 1.4159009188973233,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.0713,
      "step": 3544
    },
    {
      "epoch": 1.4163004394726328,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 3545
    },
    {
      "epoch": 1.4166999600479424,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 3546
    },
    {
      "epoch": 1.4170994806232522,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 3547
    },
    {
      "epoch": 1.4174990011985618,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005,
      "loss": 1.0715,
      "step": 3548
    },
    {
      "epoch": 1.4178985217738713,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0651,
      "step": 3549
    },
    {
      "epoch": 1.418298042349181,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 3550
    },
    {
      "epoch": 1.4186975629244907,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 3551
    },
    {
      "epoch": 1.4190970834998002,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 3552
    },
    {
      "epoch": 1.4194966040751098,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 3553
    },
    {
      "epoch": 1.4198961246504194,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 3554
    },
    {
      "epoch": 1.4202956452257292,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 3555
    },
    {
      "epoch": 1.4206951658010387,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0651,
      "step": 3556
    },
    {
      "epoch": 1.4210946863763483,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 3557
    },
    {
      "epoch": 1.421494206951658,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 3558
    },
    {
      "epoch": 1.4218937275269676,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 3559
    },
    {
      "epoch": 1.4222932481022772,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0837,
      "step": 3560
    },
    {
      "epoch": 1.422692768677587,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 3561
    },
    {
      "epoch": 1.4230922892528965,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0714,
      "step": 3562
    },
    {
      "epoch": 1.4234918098282061,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 3563
    },
    {
      "epoch": 1.423891330403516,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 3564
    },
    {
      "epoch": 1.4242908509788255,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0744,
      "step": 3565
    },
    {
      "epoch": 1.424690371554135,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 3566
    },
    {
      "epoch": 1.4250898921294446,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 3567
    },
    {
      "epoch": 1.4254894127047544,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 3568
    },
    {
      "epoch": 1.425888933280064,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 3569
    },
    {
      "epoch": 1.4262884538553735,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0714,
      "step": 3570
    },
    {
      "epoch": 1.426687974430683,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.079,
      "step": 3571
    },
    {
      "epoch": 1.4270874950059929,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 3572
    },
    {
      "epoch": 1.4274870155813024,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 3573
    },
    {
      "epoch": 1.427886536156612,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 3574
    },
    {
      "epoch": 1.4282860567319218,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 3575
    },
    {
      "epoch": 1.4286855773072313,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 3576
    },
    {
      "epoch": 1.429085097882541,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 3577
    },
    {
      "epoch": 1.4294846184578507,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 3578
    },
    {
      "epoch": 1.4298841390331603,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 3579
    },
    {
      "epoch": 1.4302836596084698,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 3580
    },
    {
      "epoch": 1.4306831801837796,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0802,
      "step": 3581
    },
    {
      "epoch": 1.4310827007590892,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 3582
    },
    {
      "epoch": 1.4314822213343987,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 3583
    },
    {
      "epoch": 1.4318817419097083,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 3584
    },
    {
      "epoch": 1.4322812624850179,
      "grad_norm": 0.265625,
      "learning_rate": 0.0005,
      "loss": 1.0822,
      "step": 3585
    },
    {
      "epoch": 1.4326807830603276,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0672,
      "step": 3586
    },
    {
      "epoch": 1.4330803036356372,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0843,
      "step": 3587
    },
    {
      "epoch": 1.4334798242109468,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 3588
    },
    {
      "epoch": 1.4338793447862566,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0716,
      "step": 3589
    },
    {
      "epoch": 1.4342788653615661,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0771,
      "step": 3590
    },
    {
      "epoch": 1.4346783859368757,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.077,
      "step": 3591
    },
    {
      "epoch": 1.4350779065121855,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0789,
      "step": 3592
    },
    {
      "epoch": 1.435477427087495,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0697,
      "step": 3593
    },
    {
      "epoch": 1.4358769476628046,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 3594
    },
    {
      "epoch": 1.4362764682381144,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 3595
    },
    {
      "epoch": 1.436675988813424,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 3596
    },
    {
      "epoch": 1.4370755093887335,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 3597
    },
    {
      "epoch": 1.437475029964043,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0811,
      "step": 3598
    },
    {
      "epoch": 1.4378745505393526,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 3599
    },
    {
      "epoch": 1.4382740711146624,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 3600
    },
    {
      "epoch": 1.438673591689972,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0693,
      "step": 3601
    },
    {
      "epoch": 1.4390731122652816,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 3602
    },
    {
      "epoch": 1.4394726328405913,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 3603
    },
    {
      "epoch": 1.439872153415901,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0719,
      "step": 3604
    },
    {
      "epoch": 1.4402716739912105,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0746,
      "step": 3605
    },
    {
      "epoch": 1.4406711945665203,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 3606
    },
    {
      "epoch": 1.4410707151418298,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 3607
    },
    {
      "epoch": 1.4414702357171394,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0763,
      "step": 3608
    },
    {
      "epoch": 1.4418697562924492,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 3609
    },
    {
      "epoch": 1.4422692768677587,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0744,
      "step": 3610
    },
    {
      "epoch": 1.4426687974430683,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 3611
    },
    {
      "epoch": 1.4430683180183779,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 3612
    },
    {
      "epoch": 1.4434678385936877,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 3613
    },
    {
      "epoch": 1.4438673591689972,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 3614
    },
    {
      "epoch": 1.4442668797443068,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.079,
      "step": 3615
    },
    {
      "epoch": 1.4446664003196164,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0715,
      "step": 3616
    },
    {
      "epoch": 1.4450659208949261,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 3617
    },
    {
      "epoch": 1.4454654414702357,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0722,
      "step": 3618
    },
    {
      "epoch": 1.4458649620455453,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 3619
    },
    {
      "epoch": 1.446264482620855,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 3620
    },
    {
      "epoch": 1.4466640031961646,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 3621
    },
    {
      "epoch": 1.4470635237714742,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0672,
      "step": 3622
    },
    {
      "epoch": 1.447463044346784,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 3623
    },
    {
      "epoch": 1.4478625649220935,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.075,
      "step": 3624
    },
    {
      "epoch": 1.448262085497403,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 3625
    },
    {
      "epoch": 1.4486616060727129,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0821,
      "step": 3626
    },
    {
      "epoch": 1.4490611266480224,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 3627
    },
    {
      "epoch": 1.449460647223332,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0797,
      "step": 3628
    },
    {
      "epoch": 1.4498601677986416,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 3629
    },
    {
      "epoch": 1.4502596883739511,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 3630
    },
    {
      "epoch": 1.450659208949261,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 3631
    },
    {
      "epoch": 1.4510587295245705,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 3632
    },
    {
      "epoch": 1.45145825009988,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 3633
    },
    {
      "epoch": 1.4518577706751898,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0712,
      "step": 3634
    },
    {
      "epoch": 1.4522572912504994,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 3635
    },
    {
      "epoch": 1.452656811825809,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.077,
      "step": 3636
    },
    {
      "epoch": 1.4530563324011188,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 3637
    },
    {
      "epoch": 1.4534558529764283,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 3638
    },
    {
      "epoch": 1.4538553735517379,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 3639
    },
    {
      "epoch": 1.4542548941270477,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 3640
    },
    {
      "epoch": 1.4546544147023572,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 3641
    },
    {
      "epoch": 1.4550539352776668,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 3642
    },
    {
      "epoch": 1.4554534558529764,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 3643
    },
    {
      "epoch": 1.455852976428286,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 3644
    },
    {
      "epoch": 1.4562524970035957,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.069,
      "step": 3645
    },
    {
      "epoch": 1.4566520175789053,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 3646
    },
    {
      "epoch": 1.4570515381542148,
      "grad_norm": 0.3125,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 3647
    },
    {
      "epoch": 1.4574510587295246,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0764,
      "step": 3648
    },
    {
      "epoch": 1.4578505793048342,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 3649
    },
    {
      "epoch": 1.4582500998801438,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 3650
    },
    {
      "epoch": 1.4586496204554535,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 3651
    },
    {
      "epoch": 1.459049141030763,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.0737,
      "step": 3652
    },
    {
      "epoch": 1.4594486616060727,
      "grad_norm": 0.21875,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 3653
    },
    {
      "epoch": 1.4598481821813825,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 3654
    },
    {
      "epoch": 1.460247702756692,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 3655
    },
    {
      "epoch": 1.4606472233320016,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 3656
    },
    {
      "epoch": 1.4610467439073112,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 3657
    },
    {
      "epoch": 1.461446264482621,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005,
      "loss": 1.0799,
      "step": 3658
    },
    {
      "epoch": 1.4618457850579305,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 3659
    },
    {
      "epoch": 1.46224530563324,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 3660
    },
    {
      "epoch": 1.4626448262085496,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0727,
      "step": 3661
    },
    {
      "epoch": 1.4630443467838594,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005,
      "loss": 1.0687,
      "step": 3662
    },
    {
      "epoch": 1.463443867359169,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 3663
    },
    {
      "epoch": 1.4638433879344785,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 3664
    },
    {
      "epoch": 1.4642429085097883,
      "grad_norm": 0.25,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 3665
    },
    {
      "epoch": 1.464642429085098,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0717,
      "step": 3666
    },
    {
      "epoch": 1.4650419496604075,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 3667
    },
    {
      "epoch": 1.4654414702357172,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 3668
    },
    {
      "epoch": 1.4658409908110268,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0697,
      "step": 3669
    },
    {
      "epoch": 1.4662405113863364,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0774,
      "step": 3670
    },
    {
      "epoch": 1.4666400319616462,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 3671
    },
    {
      "epoch": 1.4670395525369557,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 3672
    },
    {
      "epoch": 1.4674390731122653,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 3673
    },
    {
      "epoch": 1.4678385936875749,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005,
      "loss": 1.0738,
      "step": 3674
    },
    {
      "epoch": 1.4682381142628844,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0687,
      "step": 3675
    },
    {
      "epoch": 1.4686376348381942,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.0837,
      "step": 3676
    },
    {
      "epoch": 1.4690371554135038,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.077,
      "step": 3677
    },
    {
      "epoch": 1.4694366759888133,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0761,
      "step": 3678
    },
    {
      "epoch": 1.4698361965641231,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0749,
      "step": 3679
    },
    {
      "epoch": 1.4702357171394327,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0711,
      "step": 3680
    },
    {
      "epoch": 1.4706352377147422,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 3681
    },
    {
      "epoch": 1.471034758290052,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0864,
      "step": 3682
    },
    {
      "epoch": 1.4714342788653616,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.08,
      "step": 3683
    },
    {
      "epoch": 1.4718337994406712,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 3684
    },
    {
      "epoch": 1.472233320015981,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 3685
    },
    {
      "epoch": 1.4726328405912905,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 3686
    },
    {
      "epoch": 1.4730323611666,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 3687
    },
    {
      "epoch": 1.4734318817419096,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 3688
    },
    {
      "epoch": 1.4738314023172192,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 3689
    },
    {
      "epoch": 1.474230922892529,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0717,
      "step": 3690
    },
    {
      "epoch": 1.4746304434678386,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 3691
    },
    {
      "epoch": 1.4750299640431481,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0715,
      "step": 3692
    },
    {
      "epoch": 1.475429484618458,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 3693
    },
    {
      "epoch": 1.4758290051937675,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0736,
      "step": 3694
    },
    {
      "epoch": 1.476228525769077,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0715,
      "step": 3695
    },
    {
      "epoch": 1.4766280463443868,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 3696
    },
    {
      "epoch": 1.4770275669196964,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0835,
      "step": 3697
    },
    {
      "epoch": 1.477427087495006,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 3698
    },
    {
      "epoch": 1.4778266080703157,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 3699
    },
    {
      "epoch": 1.4782261286456253,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 3700
    },
    {
      "epoch": 1.4786256492209349,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0774,
      "step": 3701
    },
    {
      "epoch": 1.4790251697962444,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0715,
      "step": 3702
    },
    {
      "epoch": 1.479424690371554,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 3703
    },
    {
      "epoch": 1.4798242109468638,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 3704
    },
    {
      "epoch": 1.4802237315221733,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 3705
    },
    {
      "epoch": 1.480623252097483,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0775,
      "step": 3706
    },
    {
      "epoch": 1.4810227726727927,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0794,
      "step": 3707
    },
    {
      "epoch": 1.4814222932481023,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0793,
      "step": 3708
    },
    {
      "epoch": 1.4818218138234118,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0714,
      "step": 3709
    },
    {
      "epoch": 1.4822213343987216,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 3710
    },
    {
      "epoch": 1.4826208549740312,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 3711
    },
    {
      "epoch": 1.4830203755493407,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 3712
    },
    {
      "epoch": 1.4834198961246505,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 3713
    },
    {
      "epoch": 1.48381941669996,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0712,
      "step": 3714
    },
    {
      "epoch": 1.4842189372752697,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 3715
    },
    {
      "epoch": 1.4846184578505794,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0005,
      "loss": 1.0731,
      "step": 3716
    },
    {
      "epoch": 1.485017978425889,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 3717
    },
    {
      "epoch": 1.4854174990011986,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.0696,
      "step": 3718
    },
    {
      "epoch": 1.4858170195765081,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 3719
    },
    {
      "epoch": 1.4862165401518177,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 3720
    },
    {
      "epoch": 1.4866160607271275,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 3721
    },
    {
      "epoch": 1.487015581302437,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0696,
      "step": 3722
    },
    {
      "epoch": 1.4874151018777466,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.069,
      "step": 3723
    },
    {
      "epoch": 1.4878146224530564,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 3724
    },
    {
      "epoch": 1.488214143028366,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0774,
      "step": 3725
    },
    {
      "epoch": 1.4886136636036755,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 3726
    },
    {
      "epoch": 1.4890131841789853,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0712,
      "step": 3727
    },
    {
      "epoch": 1.4894127047542949,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 3728
    },
    {
      "epoch": 1.4898122253296044,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0657,
      "step": 3729
    },
    {
      "epoch": 1.4902117459049142,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 3730
    },
    {
      "epoch": 1.4906112664802238,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 3731
    },
    {
      "epoch": 1.4910107870555334,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 3732
    },
    {
      "epoch": 1.491410307630843,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 3733
    },
    {
      "epoch": 1.4918098282061525,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 3734
    },
    {
      "epoch": 1.4922093487814623,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0722,
      "step": 3735
    },
    {
      "epoch": 1.4926088693567718,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 3736
    },
    {
      "epoch": 1.4930083899320814,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 3737
    },
    {
      "epoch": 1.4934079105073912,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 3738
    },
    {
      "epoch": 1.4938074310827008,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0725,
      "step": 3739
    },
    {
      "epoch": 1.4942069516580103,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 3740
    },
    {
      "epoch": 1.49460647223332,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0715,
      "step": 3741
    },
    {
      "epoch": 1.4950059928086297,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0697,
      "step": 3742
    },
    {
      "epoch": 1.4954055133839392,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 3743
    },
    {
      "epoch": 1.495805033959249,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0719,
      "step": 3744
    },
    {
      "epoch": 1.4962045545345586,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 3745
    },
    {
      "epoch": 1.4966040751098681,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 3746
    },
    {
      "epoch": 1.4970035956851777,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 3747
    },
    {
      "epoch": 1.4974031162604873,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.077,
      "step": 3748
    },
    {
      "epoch": 1.497802636835797,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0665,
      "step": 3749
    },
    {
      "epoch": 1.4982021574111066,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0769,
      "step": 3750
    },
    {
      "epoch": 1.4986016779864162,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 3751
    },
    {
      "epoch": 1.499001198561726,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0735,
      "step": 3752
    },
    {
      "epoch": 1.4994007191370355,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 3753
    },
    {
      "epoch": 1.499800239712345,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0794,
      "step": 3754
    },
    {
      "epoch": 1.500199760287655,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 3755
    },
    {
      "epoch": 1.5005992808629645,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.081,
      "step": 3756
    },
    {
      "epoch": 1.500998801438274,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0672,
      "step": 3757
    },
    {
      "epoch": 1.5013983220135838,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.086,
      "step": 3758
    },
    {
      "epoch": 1.5017978425888934,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 3759
    },
    {
      "epoch": 1.502197363164203,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0776,
      "step": 3760
    },
    {
      "epoch": 1.5025968837395127,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 3761
    },
    {
      "epoch": 1.502996404314822,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 3762
    },
    {
      "epoch": 1.5033959248901319,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 3763
    },
    {
      "epoch": 1.5037954454654414,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 3764
    },
    {
      "epoch": 1.504194966040751,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 3765
    },
    {
      "epoch": 1.5045944866160608,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 3766
    },
    {
      "epoch": 1.5049940071913703,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 3767
    },
    {
      "epoch": 1.50539352776668,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 3768
    },
    {
      "epoch": 1.5057930483419897,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0734,
      "step": 3769
    },
    {
      "epoch": 1.5061925689172992,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0788,
      "step": 3770
    },
    {
      "epoch": 1.5065920894926088,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 3771
    },
    {
      "epoch": 1.5069916100679186,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0721,
      "step": 3772
    },
    {
      "epoch": 1.5073911306432282,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0725,
      "step": 3773
    },
    {
      "epoch": 1.5077906512185377,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 3774
    },
    {
      "epoch": 1.5081901717938475,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0745,
      "step": 3775
    },
    {
      "epoch": 1.5085896923691569,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 3776
    },
    {
      "epoch": 1.5089892129444666,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 3777
    },
    {
      "epoch": 1.5093887335197764,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 3778
    },
    {
      "epoch": 1.5097882540950858,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 3779
    },
    {
      "epoch": 1.5101877746703956,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 3780
    },
    {
      "epoch": 1.5105872952457051,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0759,
      "step": 3781
    },
    {
      "epoch": 1.5109868158210147,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 3782
    },
    {
      "epoch": 1.5113863363963245,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 3783
    },
    {
      "epoch": 1.511785856971634,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 3784
    },
    {
      "epoch": 1.5121853775469436,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 3785
    },
    {
      "epoch": 1.5125848981222534,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 3786
    },
    {
      "epoch": 1.512984418697563,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 3787
    },
    {
      "epoch": 1.5133839392728725,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0827,
      "step": 3788
    },
    {
      "epoch": 1.5137834598481823,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 3789
    },
    {
      "epoch": 1.5141829804234919,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 3790
    },
    {
      "epoch": 1.5145825009988014,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 3791
    },
    {
      "epoch": 1.5149820215741112,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 3792
    },
    {
      "epoch": 1.5153815421494206,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 3793
    },
    {
      "epoch": 1.5157810627247303,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0665,
      "step": 3794
    },
    {
      "epoch": 1.51618058330004,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 3795
    },
    {
      "epoch": 1.5165801038753495,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 3796
    },
    {
      "epoch": 1.5169796244506593,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 3797
    },
    {
      "epoch": 1.5173791450259688,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 3798
    },
    {
      "epoch": 1.5177786656012784,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 3799
    },
    {
      "epoch": 1.5181781861765882,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0704,
      "step": 3800
    },
    {
      "epoch": 1.5185777067518977,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 3801
    },
    {
      "epoch": 1.5189772273272073,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 3802
    },
    {
      "epoch": 1.519376747902517,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0725,
      "step": 3803
    },
    {
      "epoch": 1.5197762684778267,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0808,
      "step": 3804
    },
    {
      "epoch": 1.5201757890531362,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 3805
    },
    {
      "epoch": 1.520575309628446,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 3806
    },
    {
      "epoch": 1.5209748302037553,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.07,
      "step": 3807
    },
    {
      "epoch": 1.5213743507790651,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0005,
      "loss": 1.0783,
      "step": 3808
    },
    {
      "epoch": 1.5217738713543747,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 3809
    },
    {
      "epoch": 1.5221733919296843,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 3810
    },
    {
      "epoch": 1.522572912504994,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0774,
      "step": 3811
    },
    {
      "epoch": 1.5229724330803036,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 3812
    },
    {
      "epoch": 1.5233719536556132,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 3813
    },
    {
      "epoch": 1.523771474230923,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 3814
    },
    {
      "epoch": 1.5241709948062325,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0005,
      "loss": 1.0729,
      "step": 3815
    },
    {
      "epoch": 1.524570515381542,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0766,
      "step": 3816
    },
    {
      "epoch": 1.5249700359568519,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 3817
    },
    {
      "epoch": 1.5253695565321614,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0735,
      "step": 3818
    },
    {
      "epoch": 1.525769077107471,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 3819
    },
    {
      "epoch": 1.5261685976827808,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 3820
    },
    {
      "epoch": 1.5265681182580901,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0772,
      "step": 3821
    },
    {
      "epoch": 1.5269676388334,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005,
      "loss": 1.0821,
      "step": 3822
    },
    {
      "epoch": 1.5273671594087097,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 3823
    },
    {
      "epoch": 1.527766679984019,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 3824
    },
    {
      "epoch": 1.5281662005593288,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 3825
    },
    {
      "epoch": 1.5285657211346384,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 3826
    },
    {
      "epoch": 1.528965241709948,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 3827
    },
    {
      "epoch": 1.5293647622852578,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 3828
    },
    {
      "epoch": 1.5297642828605673,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 3829
    },
    {
      "epoch": 1.5301638034358769,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0725,
      "step": 3830
    },
    {
      "epoch": 1.5305633240111867,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0697,
      "step": 3831
    },
    {
      "epoch": 1.5309628445864962,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 3832
    },
    {
      "epoch": 1.5313623651618058,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 3833
    },
    {
      "epoch": 1.5317618857371156,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 3834
    },
    {
      "epoch": 1.5321614063124251,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 3835
    },
    {
      "epoch": 1.5325609268877347,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 3836
    },
    {
      "epoch": 1.5329604474630445,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 3837
    },
    {
      "epoch": 1.5333599680383538,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0711,
      "step": 3838
    },
    {
      "epoch": 1.5337594886136636,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 3839
    },
    {
      "epoch": 1.5341590091889732,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0753,
      "step": 3840
    },
    {
      "epoch": 1.5345585297642828,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0734,
      "step": 3841
    },
    {
      "epoch": 1.5349580503395925,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 3842
    },
    {
      "epoch": 1.535357570914902,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 3843
    },
    {
      "epoch": 1.5357570914902117,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0742,
      "step": 3844
    },
    {
      "epoch": 1.5361566120655215,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 3845
    },
    {
      "epoch": 1.536556132640831,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 3846
    },
    {
      "epoch": 1.5369556532161406,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 3847
    },
    {
      "epoch": 1.5373551737914504,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 3848
    },
    {
      "epoch": 1.53775469436676,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0779,
      "step": 3849
    },
    {
      "epoch": 1.5381542149420695,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 3850
    },
    {
      "epoch": 1.5385537355173793,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0777,
      "step": 3851
    },
    {
      "epoch": 1.5389532560926886,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 3852
    },
    {
      "epoch": 1.5393527766679984,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 3853
    },
    {
      "epoch": 1.539752297243308,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0721,
      "step": 3854
    },
    {
      "epoch": 1.5401518178186175,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 3855
    },
    {
      "epoch": 1.5405513383939273,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 3856
    },
    {
      "epoch": 1.540950858969237,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 3857
    },
    {
      "epoch": 1.5413503795445465,
      "grad_norm": 0.28125,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 3858
    },
    {
      "epoch": 1.5417499001198562,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0752,
      "step": 3859
    },
    {
      "epoch": 1.5421494206951658,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 3860
    },
    {
      "epoch": 1.5425489412704754,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 3861
    },
    {
      "epoch": 1.5429484618457852,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 3862
    },
    {
      "epoch": 1.5433479824210947,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0005,
      "loss": 1.0727,
      "step": 3863
    },
    {
      "epoch": 1.5437475029964043,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0769,
      "step": 3864
    },
    {
      "epoch": 1.544147023571714,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 3865
    },
    {
      "epoch": 1.5445465441470234,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.076,
      "step": 3866
    },
    {
      "epoch": 1.5449460647223332,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 3867
    },
    {
      "epoch": 1.545345585297643,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0773,
      "step": 3868
    },
    {
      "epoch": 1.5457451058729523,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 3869
    },
    {
      "epoch": 1.5461446264482621,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 3870
    },
    {
      "epoch": 1.5465441470235717,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 3871
    },
    {
      "epoch": 1.5469436675988812,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 3872
    },
    {
      "epoch": 1.547343188174191,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 3873
    },
    {
      "epoch": 1.5477427087495006,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0714,
      "step": 3874
    },
    {
      "epoch": 1.5481422293248102,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 3875
    },
    {
      "epoch": 1.54854174990012,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 3876
    },
    {
      "epoch": 1.5489412704754295,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 3877
    },
    {
      "epoch": 1.549340791050739,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0005,
      "loss": 1.077,
      "step": 3878
    },
    {
      "epoch": 1.5497403116260489,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 3879
    },
    {
      "epoch": 1.5501398322013584,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0819,
      "step": 3880
    },
    {
      "epoch": 1.550539352776668,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 3881
    },
    {
      "epoch": 1.5509388733519778,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 3882
    },
    {
      "epoch": 1.5513383939272871,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0757,
      "step": 3883
    },
    {
      "epoch": 1.551737914502597,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0865,
      "step": 3884
    },
    {
      "epoch": 1.5521374350779065,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 3885
    },
    {
      "epoch": 1.552536955653216,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 3886
    },
    {
      "epoch": 1.5529364762285258,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0731,
      "step": 3887
    },
    {
      "epoch": 1.5533359968038354,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005,
      "loss": 1.0762,
      "step": 3888
    },
    {
      "epoch": 1.553735517379145,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0729,
      "step": 3889
    },
    {
      "epoch": 1.5541350379544547,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 3890
    },
    {
      "epoch": 1.5545345585297643,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 3891
    },
    {
      "epoch": 1.5549340791050739,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 3892
    },
    {
      "epoch": 1.5553335996803836,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0754,
      "step": 3893
    },
    {
      "epoch": 1.5557331202556932,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0759,
      "step": 3894
    },
    {
      "epoch": 1.5561326408310028,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 3895
    },
    {
      "epoch": 1.5565321614063126,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 3896
    },
    {
      "epoch": 1.556931681981622,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 3897
    },
    {
      "epoch": 1.5573312025569317,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 3898
    },
    {
      "epoch": 1.5577307231322413,
      "grad_norm": 0.4375,
      "learning_rate": 0.0005,
      "loss": 1.0731,
      "step": 3899
    },
    {
      "epoch": 1.5581302437075508,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 3900
    },
    {
      "epoch": 1.5585297642828606,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0005,
      "loss": 1.0782,
      "step": 3901
    },
    {
      "epoch": 1.5589292848581702,
      "grad_norm": 0.640625,
      "learning_rate": 0.0005,
      "loss": 1.0717,
      "step": 3902
    },
    {
      "epoch": 1.5593288054334797,
      "grad_norm": 0.59375,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 3903
    },
    {
      "epoch": 1.5597283260087895,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 3904
    },
    {
      "epoch": 1.560127846584099,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005,
      "loss": 1.0732,
      "step": 3905
    },
    {
      "epoch": 1.5605273671594087,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0005,
      "loss": 1.0802,
      "step": 3906
    },
    {
      "epoch": 1.5609268877347184,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 3907
    },
    {
      "epoch": 1.561326408310028,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 3908
    },
    {
      "epoch": 1.5617259288853376,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005,
      "loss": 1.0736,
      "step": 3909
    },
    {
      "epoch": 1.5621254494606474,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 3910
    },
    {
      "epoch": 1.5625249700359567,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0005,
      "loss": 1.0833,
      "step": 3911
    },
    {
      "epoch": 1.5629244906112665,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0734,
      "step": 3912
    },
    {
      "epoch": 1.5633240111865763,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0005,
      "loss": 1.0817,
      "step": 3913
    },
    {
      "epoch": 1.5637235317618856,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0005,
      "loss": 1.0864,
      "step": 3914
    },
    {
      "epoch": 1.5641230523371954,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 3915
    },
    {
      "epoch": 1.564522572912505,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 3916
    },
    {
      "epoch": 1.5649220934878145,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 3917
    },
    {
      "epoch": 1.5653216140631243,
      "grad_norm": 0.5,
      "learning_rate": 0.0005,
      "loss": 1.0742,
      "step": 3918
    },
    {
      "epoch": 1.5657211346384339,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 3919
    },
    {
      "epoch": 1.5661206552137434,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0005,
      "loss": 1.0797,
      "step": 3920
    },
    {
      "epoch": 1.5665201757890532,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 3921
    },
    {
      "epoch": 1.5669196963643628,
      "grad_norm": 0.25,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 3922
    },
    {
      "epoch": 1.5673192169396724,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 3923
    },
    {
      "epoch": 1.5677187375149821,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 3924
    },
    {
      "epoch": 1.5681182580902915,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 3925
    },
    {
      "epoch": 1.5685177786656013,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 3926
    },
    {
      "epoch": 1.568917299240911,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 3927
    },
    {
      "epoch": 1.5693168198162204,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 3928
    },
    {
      "epoch": 1.5697163403915302,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0755,
      "step": 3929
    },
    {
      "epoch": 1.5701158609668397,
      "grad_norm": 0.40625,
      "learning_rate": 0.0005,
      "loss": 1.0769,
      "step": 3930
    },
    {
      "epoch": 1.5705153815421493,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 3931
    },
    {
      "epoch": 1.570914902117459,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0005,
      "loss": 1.0787,
      "step": 3932
    },
    {
      "epoch": 1.5713144226927687,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 3933
    },
    {
      "epoch": 1.5717139432680782,
      "grad_norm": 0.25,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 3934
    },
    {
      "epoch": 1.572113463843388,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 3935
    },
    {
      "epoch": 1.5725129844186976,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0005,
      "loss": 1.0811,
      "step": 3936
    },
    {
      "epoch": 1.5729125049940071,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 3937
    },
    {
      "epoch": 1.573312025569317,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 3938
    },
    {
      "epoch": 1.5737115461446265,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0746,
      "step": 3939
    },
    {
      "epoch": 1.574111066719936,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005,
      "loss": 1.0729,
      "step": 3940
    },
    {
      "epoch": 1.5745105872952458,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 3941
    },
    {
      "epoch": 1.5749101078705552,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 3942
    },
    {
      "epoch": 1.575309628445865,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0744,
      "step": 3943
    },
    {
      "epoch": 1.5757091490211745,
      "grad_norm": 0.265625,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 3944
    },
    {
      "epoch": 1.576108669596484,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0005,
      "loss": 1.0736,
      "step": 3945
    },
    {
      "epoch": 1.5765081901717939,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005,
      "loss": 1.0806,
      "step": 3946
    },
    {
      "epoch": 1.5769077107471035,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0711,
      "step": 3947
    },
    {
      "epoch": 1.577307231322413,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005,
      "loss": 1.0808,
      "step": 3948
    },
    {
      "epoch": 1.5777067518977228,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0787,
      "step": 3949
    },
    {
      "epoch": 1.5781062724730324,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 3950
    },
    {
      "epoch": 1.578505793048342,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 3951
    },
    {
      "epoch": 1.5789053136236517,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 3952
    },
    {
      "epoch": 1.5793048341989613,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 3953
    },
    {
      "epoch": 1.5797043547742708,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 3954
    },
    {
      "epoch": 1.5801038753495806,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0005,
      "loss": 1.0719,
      "step": 3955
    },
    {
      "epoch": 1.58050339592489,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0005,
      "loss": 1.0754,
      "step": 3956
    },
    {
      "epoch": 1.5809029165001998,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 3957
    },
    {
      "epoch": 1.5813024370755095,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0005,
      "loss": 1.0713,
      "step": 3958
    },
    {
      "epoch": 1.581701957650819,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 3959
    },
    {
      "epoch": 1.5821014782261287,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 3960
    },
    {
      "epoch": 1.5825009988014382,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0732,
      "step": 3961
    },
    {
      "epoch": 1.5829005193767478,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 3962
    },
    {
      "epoch": 1.5833000399520576,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 3963
    },
    {
      "epoch": 1.5836995605273672,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0687,
      "step": 3964
    },
    {
      "epoch": 1.5840990811026767,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 3965
    },
    {
      "epoch": 1.5844986016779865,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 3966
    },
    {
      "epoch": 1.584898122253296,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 3967
    },
    {
      "epoch": 1.5852976428286056,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 3968
    },
    {
      "epoch": 1.5856971634039154,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 3969
    },
    {
      "epoch": 1.5860966839792248,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 3970
    },
    {
      "epoch": 1.5864962045545346,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 3971
    },
    {
      "epoch": 1.5868957251298443,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0679,
      "step": 3972
    },
    {
      "epoch": 1.5872952457051537,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 3973
    },
    {
      "epoch": 1.5876947662804635,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0768,
      "step": 3974
    },
    {
      "epoch": 1.588094286855773,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0005,
      "loss": 1.0746,
      "step": 3975
    },
    {
      "epoch": 1.5884938074310826,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 3976
    },
    {
      "epoch": 1.5888933280063924,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 3977
    },
    {
      "epoch": 1.589292848581702,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0005,
      "loss": 1.0728,
      "step": 3978
    },
    {
      "epoch": 1.5896923691570115,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 3979
    },
    {
      "epoch": 1.5900918897323213,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0727,
      "step": 3980
    },
    {
      "epoch": 1.5904914103076309,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0005,
      "loss": 1.0817,
      "step": 3981
    },
    {
      "epoch": 1.5908909308829404,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 3982
    },
    {
      "epoch": 1.5912904514582502,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0005,
      "loss": 1.076,
      "step": 3983
    },
    {
      "epoch": 1.5916899720335598,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 3984
    },
    {
      "epoch": 1.5920894926088693,
      "grad_norm": 0.609375,
      "learning_rate": 0.0005,
      "loss": 1.0814,
      "step": 3985
    },
    {
      "epoch": 1.5924890131841791,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0005,
      "loss": 1.0712,
      "step": 3986
    },
    {
      "epoch": 1.5928885337594885,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 3987
    },
    {
      "epoch": 1.5932880543347983,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0798,
      "step": 3988
    },
    {
      "epoch": 1.5936875749101078,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0697,
      "step": 3989
    },
    {
      "epoch": 1.5940870954854174,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0763,
      "step": 3990
    },
    {
      "epoch": 1.5944866160607272,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 3991
    },
    {
      "epoch": 1.5948861366360367,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 3992
    },
    {
      "epoch": 1.5952856572113463,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 3993
    },
    {
      "epoch": 1.595685177786656,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 3994
    },
    {
      "epoch": 1.5960846983619656,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0005,
      "loss": 1.0779,
      "step": 3995
    },
    {
      "epoch": 1.5964842189372752,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 3996
    },
    {
      "epoch": 1.596883739512585,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.08,
      "step": 3997
    },
    {
      "epoch": 1.5972832600878946,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 3998
    },
    {
      "epoch": 1.5976827806632041,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0679,
      "step": 3999
    },
    {
      "epoch": 1.598082301238514,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 4000
    },
    {
      "epoch": 1.5984818218138233,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 4001
    },
    {
      "epoch": 1.598881342389133,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 4002
    },
    {
      "epoch": 1.5992808629644428,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 4003
    },
    {
      "epoch": 1.5996803835397522,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0704,
      "step": 4004
    },
    {
      "epoch": 1.600079904115062,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 4005
    },
    {
      "epoch": 1.6004794246903715,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 4006
    },
    {
      "epoch": 1.600878945265681,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 4007
    },
    {
      "epoch": 1.6012784658409909,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 4008
    },
    {
      "epoch": 1.6016779864163004,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 4009
    },
    {
      "epoch": 1.60207750699161,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 4010
    },
    {
      "epoch": 1.6024770275669198,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 4011
    },
    {
      "epoch": 1.6028765481422294,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 4012
    },
    {
      "epoch": 1.603276068717539,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.079,
      "step": 4013
    },
    {
      "epoch": 1.6036755892928487,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 4014
    },
    {
      "epoch": 1.604075109868158,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0812,
      "step": 4015
    },
    {
      "epoch": 1.6044746304434678,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 4016
    },
    {
      "epoch": 1.6048741510187776,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 4017
    },
    {
      "epoch": 1.605273671594087,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 4018
    },
    {
      "epoch": 1.6056731921693967,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 4019
    },
    {
      "epoch": 1.6060727127447063,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 4020
    },
    {
      "epoch": 1.6064722333200159,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 4021
    },
    {
      "epoch": 1.6068717538953257,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0679,
      "step": 4022
    },
    {
      "epoch": 1.6072712744706352,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0752,
      "step": 4023
    },
    {
      "epoch": 1.6076707950459448,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 4024
    },
    {
      "epoch": 1.6080703156212546,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0657,
      "step": 4025
    },
    {
      "epoch": 1.6084698361965641,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 4026
    },
    {
      "epoch": 1.6088693567718737,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0744,
      "step": 4027
    },
    {
      "epoch": 1.6092688773471835,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 4028
    },
    {
      "epoch": 1.609668397922493,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 4029
    },
    {
      "epoch": 1.6100679184978026,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 4030
    },
    {
      "epoch": 1.6104674390731124,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0806,
      "step": 4031
    },
    {
      "epoch": 1.6108669596484217,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 4032
    },
    {
      "epoch": 1.6112664802237315,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 4033
    },
    {
      "epoch": 1.611666000799041,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 4034
    },
    {
      "epoch": 1.6120655213743507,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0786,
      "step": 4035
    },
    {
      "epoch": 1.6124650419496604,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 4036
    },
    {
      "epoch": 1.61286456252497,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 4037
    },
    {
      "epoch": 1.6132640831002796,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0842,
      "step": 4038
    },
    {
      "epoch": 1.6136636036755894,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 4039
    },
    {
      "epoch": 1.614063124250899,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0818,
      "step": 4040
    },
    {
      "epoch": 1.6144626448262085,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.08,
      "step": 4041
    },
    {
      "epoch": 1.6148621654015183,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 4042
    },
    {
      "epoch": 1.6152616859768278,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 4043
    },
    {
      "epoch": 1.6156612065521374,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 4044
    },
    {
      "epoch": 1.6160607271274472,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 4045
    },
    {
      "epoch": 1.6164602477027565,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 4046
    },
    {
      "epoch": 1.6168597682780663,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 4047
    },
    {
      "epoch": 1.617259288853376,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 4048
    },
    {
      "epoch": 1.6176588094286855,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 4049
    },
    {
      "epoch": 1.6180583300039952,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 4050
    },
    {
      "epoch": 1.6184578505793048,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 4051
    },
    {
      "epoch": 1.6188573711546144,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 4052
    },
    {
      "epoch": 1.6192568917299242,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.074,
      "step": 4053
    },
    {
      "epoch": 1.6196564123052337,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 4054
    },
    {
      "epoch": 1.6200559328805433,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 4055
    },
    {
      "epoch": 1.620455453455853,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 4056
    },
    {
      "epoch": 1.6208549740311626,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 4057
    },
    {
      "epoch": 1.6212544946064722,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0749,
      "step": 4058
    },
    {
      "epoch": 1.621654015181782,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 4059
    },
    {
      "epoch": 1.6220535357570913,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 4060
    },
    {
      "epoch": 1.6224530563324011,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 4061
    },
    {
      "epoch": 1.622852576907711,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 4062
    },
    {
      "epoch": 1.6232520974830202,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 4063
    },
    {
      "epoch": 1.62365161805833,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 4064
    },
    {
      "epoch": 1.6240511386336396,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 4065
    },
    {
      "epoch": 1.6244506592089492,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 4066
    },
    {
      "epoch": 1.624850179784259,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 4067
    },
    {
      "epoch": 1.6252497003595685,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 4068
    },
    {
      "epoch": 1.625649220934878,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0741,
      "step": 4069
    },
    {
      "epoch": 1.6260487415101879,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 4070
    },
    {
      "epoch": 1.6264482620854974,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 4071
    },
    {
      "epoch": 1.626847782660807,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0706,
      "step": 4072
    },
    {
      "epoch": 1.6272473032361168,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 4073
    },
    {
      "epoch": 1.6276468238114263,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 4074
    },
    {
      "epoch": 1.628046344386736,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.078,
      "step": 4075
    },
    {
      "epoch": 1.6284458649620457,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0696,
      "step": 4076
    },
    {
      "epoch": 1.628845385537355,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 4077
    },
    {
      "epoch": 1.6292449061126648,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.082,
      "step": 4078
    },
    {
      "epoch": 1.6296444266879744,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 4079
    },
    {
      "epoch": 1.630043947263284,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 4080
    },
    {
      "epoch": 1.6304434678385937,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 4081
    },
    {
      "epoch": 1.6308429884139033,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 4082
    },
    {
      "epoch": 1.6312425089892129,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 4083
    },
    {
      "epoch": 1.6316420295645226,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 4084
    },
    {
      "epoch": 1.6320415501398322,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 4085
    },
    {
      "epoch": 1.6324410707151418,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 4086
    },
    {
      "epoch": 1.6328405912904516,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 4087
    },
    {
      "epoch": 1.6332401118657611,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0005,
      "loss": 1.0716,
      "step": 4088
    },
    {
      "epoch": 1.6336396324410707,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 4089
    },
    {
      "epoch": 1.6340391530163805,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 4090
    },
    {
      "epoch": 1.6344386735916898,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.073,
      "step": 4091
    },
    {
      "epoch": 1.6348381941669996,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 4092
    },
    {
      "epoch": 1.6352377147423094,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 4093
    },
    {
      "epoch": 1.6356372353176187,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 4094
    },
    {
      "epoch": 1.6360367558929285,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 4095
    },
    {
      "epoch": 1.636436276468238,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 4096
    },
    {
      "epoch": 1.6368357970435476,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0842,
      "step": 4097
    },
    {
      "epoch": 1.6372353176188574,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 4098
    },
    {
      "epoch": 1.637634838194167,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 4099
    },
    {
      "epoch": 1.6380343587694766,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 4100
    },
    {
      "epoch": 1.6384338793447863,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 4101
    },
    {
      "epoch": 1.638833399920096,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 4102
    },
    {
      "epoch": 1.6392329204954055,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0775,
      "step": 4103
    },
    {
      "epoch": 1.6396324410707153,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 4104
    },
    {
      "epoch": 1.6400319616460246,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 4105
    },
    {
      "epoch": 1.6404314822213344,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 4106
    },
    {
      "epoch": 1.6408310027966442,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 4107
    },
    {
      "epoch": 1.6412305233719535,
      "grad_norm": 0.3125,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 4108
    },
    {
      "epoch": 1.6416300439472633,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 4109
    },
    {
      "epoch": 1.6420295645225729,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 4110
    },
    {
      "epoch": 1.6424290850978824,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 4111
    },
    {
      "epoch": 1.6428286056731922,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0005,
      "loss": 1.0752,
      "step": 4112
    },
    {
      "epoch": 1.6432281262485018,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 4113
    },
    {
      "epoch": 1.6436276468238114,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0005,
      "loss": 1.087,
      "step": 4114
    },
    {
      "epoch": 1.6440271673991211,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0005,
      "loss": 1.0814,
      "step": 4115
    },
    {
      "epoch": 1.6444266879744307,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 4116
    },
    {
      "epoch": 1.6448262085497403,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0706,
      "step": 4117
    },
    {
      "epoch": 1.64522572912505,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 4118
    },
    {
      "epoch": 1.6456252497003596,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0768,
      "step": 4119
    },
    {
      "epoch": 1.6460247702756692,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 4120
    },
    {
      "epoch": 1.646424290850979,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 4121
    },
    {
      "epoch": 1.6468238114262883,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 4122
    },
    {
      "epoch": 1.647223332001598,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 4123
    },
    {
      "epoch": 1.6476228525769077,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.0787,
      "step": 4124
    },
    {
      "epoch": 1.6480223731522172,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 4125
    },
    {
      "epoch": 1.648421893727527,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 4126
    },
    {
      "epoch": 1.6488214143028366,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 4127
    },
    {
      "epoch": 1.6492209348781461,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 4128
    },
    {
      "epoch": 1.649620455453456,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 4129
    },
    {
      "epoch": 1.6500199760287655,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0005,
      "loss": 1.0743,
      "step": 4130
    },
    {
      "epoch": 1.650419496604075,
      "grad_norm": 0.234375,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 4131
    },
    {
      "epoch": 1.6508190171793848,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 4132
    },
    {
      "epoch": 1.6512185377546944,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 4133
    },
    {
      "epoch": 1.651618058330004,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0005,
      "loss": 1.0711,
      "step": 4134
    },
    {
      "epoch": 1.6520175789053138,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0739,
      "step": 4135
    },
    {
      "epoch": 1.652417099480623,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0005,
      "loss": 1.0903,
      "step": 4136
    },
    {
      "epoch": 1.6528166200559329,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 4137
    },
    {
      "epoch": 1.6532161406312427,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 4138
    },
    {
      "epoch": 1.653615661206552,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 4139
    },
    {
      "epoch": 1.6540151817818618,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0005,
      "loss": 1.0809,
      "step": 4140
    },
    {
      "epoch": 1.6544147023571714,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.0738,
      "step": 4141
    },
    {
      "epoch": 1.654814222932481,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 4142
    },
    {
      "epoch": 1.6552137435077907,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 4143
    },
    {
      "epoch": 1.6556132640831003,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.0773,
      "step": 4144
    },
    {
      "epoch": 1.6560127846584098,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 4145
    },
    {
      "epoch": 1.6564123052337196,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.074,
      "step": 4146
    },
    {
      "epoch": 1.6568118258090292,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 4147
    },
    {
      "epoch": 1.6572113463843388,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 4148
    },
    {
      "epoch": 1.6576108669596485,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 4149
    },
    {
      "epoch": 1.6580103875349579,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 4150
    },
    {
      "epoch": 1.6584099081102677,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0714,
      "step": 4151
    },
    {
      "epoch": 1.6588094286855775,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 4152
    },
    {
      "epoch": 1.6592089492608868,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0745,
      "step": 4153
    },
    {
      "epoch": 1.6596084698361966,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 4154
    },
    {
      "epoch": 1.6600079904115062,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0817,
      "step": 4155
    },
    {
      "epoch": 1.6604075109868157,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 4156
    },
    {
      "epoch": 1.6608070315621255,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0728,
      "step": 4157
    },
    {
      "epoch": 1.661206552137435,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 4158
    },
    {
      "epoch": 1.6616060727127446,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 4159
    },
    {
      "epoch": 1.6620055932880544,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 4160
    },
    {
      "epoch": 1.662405113863364,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 4161
    },
    {
      "epoch": 1.6628046344386735,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0746,
      "step": 4162
    },
    {
      "epoch": 1.6632041550139833,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 4163
    },
    {
      "epoch": 1.663603675589293,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0722,
      "step": 4164
    },
    {
      "epoch": 1.6640031961646025,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 4165
    },
    {
      "epoch": 1.6644027167399122,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 4166
    },
    {
      "epoch": 1.6648022373152216,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 4167
    },
    {
      "epoch": 1.6652017578905314,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0679,
      "step": 4168
    },
    {
      "epoch": 1.665601278465841,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 4169
    },
    {
      "epoch": 1.6660007990411505,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 4170
    },
    {
      "epoch": 1.6664003196164603,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 4171
    },
    {
      "epoch": 1.6667998401917699,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 4172
    },
    {
      "epoch": 1.6671993607670794,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 4173
    },
    {
      "epoch": 1.6675988813423892,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 4174
    },
    {
      "epoch": 1.6679984019176988,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 4175
    },
    {
      "epoch": 1.6683979224930083,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 4176
    },
    {
      "epoch": 1.6687974430683181,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 4177
    },
    {
      "epoch": 1.6691969636436277,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 4178
    },
    {
      "epoch": 1.6695964842189372,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 4179
    },
    {
      "epoch": 1.669996004794247,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0679,
      "step": 4180
    },
    {
      "epoch": 1.6703955253695564,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 4181
    },
    {
      "epoch": 1.6707950459448662,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0767,
      "step": 4182
    },
    {
      "epoch": 1.671194566520176,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 4183
    },
    {
      "epoch": 1.6715940870954853,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 4184
    },
    {
      "epoch": 1.671993607670795,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 4185
    },
    {
      "epoch": 1.6723931282461046,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.076,
      "step": 4186
    },
    {
      "epoch": 1.6727926488214142,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 4187
    },
    {
      "epoch": 1.673192169396724,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 4188
    },
    {
      "epoch": 1.6735916899720336,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 4189
    },
    {
      "epoch": 1.6739912105473431,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 4190
    },
    {
      "epoch": 1.674390731122653,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0787,
      "step": 4191
    },
    {
      "epoch": 1.6747902516979625,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 4192
    },
    {
      "epoch": 1.675189772273272,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 4193
    },
    {
      "epoch": 1.6755892928485818,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 4194
    },
    {
      "epoch": 1.6759888134238912,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 4195
    },
    {
      "epoch": 1.676388333999201,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0728,
      "step": 4196
    },
    {
      "epoch": 1.6767878545745107,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 4197
    },
    {
      "epoch": 1.67718737514982,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0743,
      "step": 4198
    },
    {
      "epoch": 1.6775868957251299,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 4199
    },
    {
      "epoch": 1.6779864163004394,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 4200
    },
    {
      "epoch": 1.678385936875749,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0769,
      "step": 4201
    },
    {
      "epoch": 1.6787854574510588,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0735,
      "step": 4202
    },
    {
      "epoch": 1.6791849780263683,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 4203
    },
    {
      "epoch": 1.679584498601678,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0798,
      "step": 4204
    },
    {
      "epoch": 1.6799840191769877,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 4205
    },
    {
      "epoch": 1.6803835397522973,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 4206
    },
    {
      "epoch": 1.6807830603276068,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 4207
    },
    {
      "epoch": 1.6811825809029166,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 4208
    },
    {
      "epoch": 1.6815821014782262,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 4209
    },
    {
      "epoch": 1.6819816220535357,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 4210
    },
    {
      "epoch": 1.6823811426288455,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 4211
    },
    {
      "epoch": 1.6827806632041549,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 4212
    },
    {
      "epoch": 1.6831801837794647,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0738,
      "step": 4213
    },
    {
      "epoch": 1.6835797043547742,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 4214
    },
    {
      "epoch": 1.6839792249300838,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0865,
      "step": 4215
    },
    {
      "epoch": 1.6843787455053936,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 4216
    },
    {
      "epoch": 1.6847782660807031,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0715,
      "step": 4217
    },
    {
      "epoch": 1.6851777866560127,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 4218
    },
    {
      "epoch": 1.6855773072313225,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 4219
    },
    {
      "epoch": 1.685976827806632,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.078,
      "step": 4220
    },
    {
      "epoch": 1.6863763483819416,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 4221
    },
    {
      "epoch": 1.6867758689572514,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0804,
      "step": 4222
    },
    {
      "epoch": 1.687175389532561,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0665,
      "step": 4223
    },
    {
      "epoch": 1.6875749101078705,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 4224
    },
    {
      "epoch": 1.6879744306831803,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.075,
      "step": 4225
    },
    {
      "epoch": 1.6883739512584897,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 4226
    },
    {
      "epoch": 1.6887734718337994,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 4227
    },
    {
      "epoch": 1.6891729924091092,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0746,
      "step": 4228
    },
    {
      "epoch": 1.6895725129844186,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0665,
      "step": 4229
    },
    {
      "epoch": 1.6899720335597284,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 4230
    },
    {
      "epoch": 1.690371554135038,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 4231
    },
    {
      "epoch": 1.6907710747103475,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 4232
    },
    {
      "epoch": 1.6911705952856573,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 4233
    },
    {
      "epoch": 1.6915701158609668,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 4234
    },
    {
      "epoch": 1.6919696364362764,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 4235
    },
    {
      "epoch": 1.6923691570115862,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 4236
    },
    {
      "epoch": 1.6927686775868958,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 4237
    },
    {
      "epoch": 1.6931681981622053,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0836,
      "step": 4238
    },
    {
      "epoch": 1.693567718737515,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 4239
    },
    {
      "epoch": 1.6939672393128244,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 4240
    },
    {
      "epoch": 1.6943667598881342,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.075,
      "step": 4241
    },
    {
      "epoch": 1.694766280463444,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0737,
      "step": 4242
    },
    {
      "epoch": 1.6951658010387534,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 4243
    },
    {
      "epoch": 1.6955653216140631,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 4244
    },
    {
      "epoch": 1.6959648421893727,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 4245
    },
    {
      "epoch": 1.6963643627646823,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 4246
    },
    {
      "epoch": 1.696763883339992,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.07,
      "step": 4247
    },
    {
      "epoch": 1.6971634039153016,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 4248
    },
    {
      "epoch": 1.6975629244906112,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 4249
    },
    {
      "epoch": 1.697962445065921,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 4250
    },
    {
      "epoch": 1.6983619656412305,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 4251
    },
    {
      "epoch": 1.69876148621654,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 4252
    },
    {
      "epoch": 1.69916100679185,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 4253
    },
    {
      "epoch": 1.6995605273671595,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0672,
      "step": 4254
    },
    {
      "epoch": 1.699960047942469,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0768,
      "step": 4255
    },
    {
      "epoch": 1.7003595685177788,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 4256
    },
    {
      "epoch": 1.7007590890930881,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 4257
    },
    {
      "epoch": 1.701158609668398,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 4258
    },
    {
      "epoch": 1.7015581302437075,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0767,
      "step": 4259
    },
    {
      "epoch": 1.701957650819017,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 4260
    },
    {
      "epoch": 1.7023571713943269,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 4261
    },
    {
      "epoch": 1.7027566919696364,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0769,
      "step": 4262
    },
    {
      "epoch": 1.703156212544946,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 4263
    },
    {
      "epoch": 1.7035557331202558,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 4264
    },
    {
      "epoch": 1.7039552536955653,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0672,
      "step": 4265
    },
    {
      "epoch": 1.704354774270875,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 4266
    },
    {
      "epoch": 1.7047542948461847,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 4267
    },
    {
      "epoch": 1.7051538154214942,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 4268
    },
    {
      "epoch": 1.7055533359968038,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 4269
    },
    {
      "epoch": 1.7059528565721136,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 4270
    },
    {
      "epoch": 1.706352377147423,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 4271
    },
    {
      "epoch": 1.7067518977227327,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 4272
    },
    {
      "epoch": 1.7071514182980425,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 4273
    },
    {
      "epoch": 1.7075509388733519,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0665,
      "step": 4274
    },
    {
      "epoch": 1.7079504594486616,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 4275
    },
    {
      "epoch": 1.7083499800239712,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0803,
      "step": 4276
    },
    {
      "epoch": 1.7087495005992808,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 4277
    },
    {
      "epoch": 1.7091490211745906,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0735,
      "step": 4278
    },
    {
      "epoch": 1.7095485417499001,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 4279
    },
    {
      "epoch": 1.7099480623252097,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0762,
      "step": 4280
    },
    {
      "epoch": 1.7103475829005195,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 4281
    },
    {
      "epoch": 1.710747103475829,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 4282
    },
    {
      "epoch": 1.7111466240511386,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 4283
    },
    {
      "epoch": 1.7115461446264484,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 4284
    },
    {
      "epoch": 1.7119456652017577,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 4285
    },
    {
      "epoch": 1.7123451857770675,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 4286
    },
    {
      "epoch": 1.7127447063523773,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 4287
    },
    {
      "epoch": 1.7131442269276866,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0773,
      "step": 4288
    },
    {
      "epoch": 1.7135437475029964,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 4289
    },
    {
      "epoch": 1.713943268078306,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0734,
      "step": 4290
    },
    {
      "epoch": 1.7143427886536156,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 4291
    },
    {
      "epoch": 1.7147423092289253,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 4292
    },
    {
      "epoch": 1.715141829804235,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 4293
    },
    {
      "epoch": 1.7155413503795445,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 4294
    },
    {
      "epoch": 1.7159408709548543,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 4295
    },
    {
      "epoch": 1.7163403915301638,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 4296
    },
    {
      "epoch": 1.7167399121054734,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 4297
    },
    {
      "epoch": 1.7171394326807832,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 4298
    },
    {
      "epoch": 1.7175389532560927,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 4299
    },
    {
      "epoch": 1.7179384738314023,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 4300
    },
    {
      "epoch": 1.718337994406712,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0706,
      "step": 4301
    },
    {
      "epoch": 1.7187375149820214,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 4302
    },
    {
      "epoch": 1.7191370355573312,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 4303
    },
    {
      "epoch": 1.7195365561326408,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 4304
    },
    {
      "epoch": 1.7199360767079503,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 4305
    },
    {
      "epoch": 1.7203355972832601,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 4306
    },
    {
      "epoch": 1.7207351178585697,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 4307
    },
    {
      "epoch": 1.7211346384338793,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 4308
    },
    {
      "epoch": 1.721534159009189,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 4309
    },
    {
      "epoch": 1.7219336795844986,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 4310
    },
    {
      "epoch": 1.7223332001598082,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.07,
      "step": 4311
    },
    {
      "epoch": 1.722732720735118,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 4312
    },
    {
      "epoch": 1.7231322413104275,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 4313
    },
    {
      "epoch": 1.723531761885737,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.082,
      "step": 4314
    },
    {
      "epoch": 1.7239312824610469,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 4315
    },
    {
      "epoch": 1.7243308030363562,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0767,
      "step": 4316
    },
    {
      "epoch": 1.724730323611666,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 4317
    },
    {
      "epoch": 1.7251298441869758,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 4318
    },
    {
      "epoch": 1.7255293647622851,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 4319
    },
    {
      "epoch": 1.725928885337595,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0704,
      "step": 4320
    },
    {
      "epoch": 1.7263284059129045,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 4321
    },
    {
      "epoch": 1.726727926488214,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 4322
    },
    {
      "epoch": 1.7271274470635238,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 4323
    },
    {
      "epoch": 1.7275269676388334,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 4324
    },
    {
      "epoch": 1.727926488214143,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 4325
    },
    {
      "epoch": 1.7283260087894528,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 4326
    },
    {
      "epoch": 1.7287255293647623,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 4327
    },
    {
      "epoch": 1.7291250499400719,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 4328
    },
    {
      "epoch": 1.7295245705153817,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 4329
    },
    {
      "epoch": 1.729924091090691,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 4330
    },
    {
      "epoch": 1.7303236116660008,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 4331
    },
    {
      "epoch": 1.7307231322413106,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 4332
    },
    {
      "epoch": 1.73112265281662,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 4333
    },
    {
      "epoch": 1.7315221733919297,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 4334
    },
    {
      "epoch": 1.7319216939672393,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 4335
    },
    {
      "epoch": 1.7323212145425488,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 4336
    },
    {
      "epoch": 1.7327207351178586,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 4337
    },
    {
      "epoch": 1.7331202556931682,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 4338
    },
    {
      "epoch": 1.7335197762684778,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 4339
    },
    {
      "epoch": 1.7339192968437875,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 4340
    },
    {
      "epoch": 1.734318817419097,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 4341
    },
    {
      "epoch": 1.7347183379944067,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0735,
      "step": 4342
    },
    {
      "epoch": 1.7351178585697165,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0714,
      "step": 4343
    },
    {
      "epoch": 1.735517379145026,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 4344
    },
    {
      "epoch": 1.7359168997203356,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 4345
    },
    {
      "epoch": 1.7363164202956454,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0697,
      "step": 4346
    },
    {
      "epoch": 1.7367159408709547,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 4347
    },
    {
      "epoch": 1.7371154614462645,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 4348
    },
    {
      "epoch": 1.737514982021574,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 4349
    },
    {
      "epoch": 1.7379145025968836,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 4350
    },
    {
      "epoch": 1.7383140231721934,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 4351
    },
    {
      "epoch": 1.738713543747503,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 4352
    },
    {
      "epoch": 1.7391130643228125,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 4353
    },
    {
      "epoch": 1.7395125848981223,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0729,
      "step": 4354
    },
    {
      "epoch": 1.739912105473432,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 4355
    },
    {
      "epoch": 1.7403116260487415,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 4356
    },
    {
      "epoch": 1.7407111466240512,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.08,
      "step": 4357
    },
    {
      "epoch": 1.7411106671993608,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 4358
    },
    {
      "epoch": 1.7415101877746704,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 4359
    },
    {
      "epoch": 1.7419097083499802,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 4360
    },
    {
      "epoch": 1.7423092289252895,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 4361
    },
    {
      "epoch": 1.7427087495005993,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 4362
    },
    {
      "epoch": 1.743108270075909,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 4363
    },
    {
      "epoch": 1.7435077906512184,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 4364
    },
    {
      "epoch": 1.7439073112265282,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0819,
      "step": 4365
    },
    {
      "epoch": 1.7443068318018378,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 4366
    },
    {
      "epoch": 1.7447063523771473,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 4367
    },
    {
      "epoch": 1.7451058729524571,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 4368
    },
    {
      "epoch": 1.7455053935277667,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 4369
    },
    {
      "epoch": 1.7459049141030762,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 4370
    },
    {
      "epoch": 1.746304434678386,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 4371
    },
    {
      "epoch": 1.7467039552536956,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 4372
    },
    {
      "epoch": 1.7471034758290052,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 4373
    },
    {
      "epoch": 1.747502996404315,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 4374
    },
    {
      "epoch": 1.7479025169796243,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 4375
    },
    {
      "epoch": 1.748302037554934,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 4376
    },
    {
      "epoch": 1.7487015581302439,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 4377
    },
    {
      "epoch": 1.7491010787055532,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 4378
    },
    {
      "epoch": 1.749500599280863,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0696,
      "step": 4379
    },
    {
      "epoch": 1.7499001198561726,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 4380
    },
    {
      "epoch": 1.7502996404314821,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 4381
    },
    {
      "epoch": 1.750699161006792,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 4382
    },
    {
      "epoch": 1.7510986815821015,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0598,
      "step": 4383
    },
    {
      "epoch": 1.751498202157411,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0744,
      "step": 4384
    },
    {
      "epoch": 1.7518977227327208,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0714,
      "step": 4385
    },
    {
      "epoch": 1.7522972433080304,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 4386
    },
    {
      "epoch": 1.75269676388334,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 4387
    },
    {
      "epoch": 1.7530962844586497,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 4388
    },
    {
      "epoch": 1.7534958050339593,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0784,
      "step": 4389
    },
    {
      "epoch": 1.7538953256092689,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 4390
    },
    {
      "epoch": 1.7542948461845786,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 4391
    },
    {
      "epoch": 1.754694366759888,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 4392
    },
    {
      "epoch": 1.7550938873351978,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 4393
    },
    {
      "epoch": 1.7554934079105073,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 4394
    },
    {
      "epoch": 1.755892928485817,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 4395
    },
    {
      "epoch": 1.7562924490611267,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 4396
    },
    {
      "epoch": 1.7566919696364363,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0672,
      "step": 4397
    },
    {
      "epoch": 1.7570914902117458,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 4398
    },
    {
      "epoch": 1.7574910107870556,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 4399
    },
    {
      "epoch": 1.7578905313623652,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 4400
    },
    {
      "epoch": 1.7582900519376747,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 4401
    },
    {
      "epoch": 1.7586895725129845,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 4402
    },
    {
      "epoch": 1.759089093088294,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 4403
    },
    {
      "epoch": 1.7594886136636037,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 4404
    },
    {
      "epoch": 1.7598881342389134,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 4405
    },
    {
      "epoch": 1.7602876548142228,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 4406
    },
    {
      "epoch": 1.7606871753895326,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 4407
    },
    {
      "epoch": 1.7610866959648424,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 4408
    },
    {
      "epoch": 1.7614862165401517,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0719,
      "step": 4409
    },
    {
      "epoch": 1.7618857371154615,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0696,
      "step": 4410
    },
    {
      "epoch": 1.762285257690771,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 4411
    },
    {
      "epoch": 1.7626847782660806,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 4412
    },
    {
      "epoch": 1.7630842988413904,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 4413
    },
    {
      "epoch": 1.7634838194167,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0749,
      "step": 4414
    },
    {
      "epoch": 1.7638833399920095,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 4415
    },
    {
      "epoch": 1.7642828605673193,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 4416
    },
    {
      "epoch": 1.7646823811426289,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 4417
    },
    {
      "epoch": 1.7650819017179384,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 4418
    },
    {
      "epoch": 1.7654814222932482,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 4419
    },
    {
      "epoch": 1.7658809428685576,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 4420
    },
    {
      "epoch": 1.7662804634438674,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0672,
      "step": 4421
    },
    {
      "epoch": 1.7666799840191771,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 4422
    },
    {
      "epoch": 1.7670795045944865,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 4423
    },
    {
      "epoch": 1.7674790251697963,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 4424
    },
    {
      "epoch": 1.7678785457451058,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 4425
    },
    {
      "epoch": 1.7682780663204154,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 4426
    },
    {
      "epoch": 1.7686775868957252,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 4427
    },
    {
      "epoch": 1.7690771074710347,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 4428
    },
    {
      "epoch": 1.7694766280463443,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 4429
    },
    {
      "epoch": 1.769876148621654,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 4430
    },
    {
      "epoch": 1.7702756691969637,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 4431
    },
    {
      "epoch": 1.7706751897722732,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 4432
    },
    {
      "epoch": 1.771074710347583,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0735,
      "step": 4433
    },
    {
      "epoch": 1.7714742309228926,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 4434
    },
    {
      "epoch": 1.7718737514982021,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 4435
    },
    {
      "epoch": 1.772273272073512,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0665,
      "step": 4436
    },
    {
      "epoch": 1.7726727926488213,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 4437
    },
    {
      "epoch": 1.773072313224131,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 4438
    },
    {
      "epoch": 1.7734718337994406,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 4439
    },
    {
      "epoch": 1.7738713543747502,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 4440
    },
    {
      "epoch": 1.77427087495006,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 4441
    },
    {
      "epoch": 1.7746703955253695,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 4442
    },
    {
      "epoch": 1.775069916100679,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 4443
    },
    {
      "epoch": 1.7754694366759889,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0696,
      "step": 4444
    },
    {
      "epoch": 1.7758689572512985,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0713,
      "step": 4445
    },
    {
      "epoch": 1.776268477826608,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0732,
      "step": 4446
    },
    {
      "epoch": 1.7766679984019178,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0739,
      "step": 4447
    },
    {
      "epoch": 1.7770675189772274,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 4448
    },
    {
      "epoch": 1.777467039552537,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 4449
    },
    {
      "epoch": 1.7778665601278467,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 4450
    },
    {
      "epoch": 1.778266080703156,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 4451
    },
    {
      "epoch": 1.7786656012784658,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 4452
    },
    {
      "epoch": 1.7790651218537756,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0679,
      "step": 4453
    },
    {
      "epoch": 1.779464642429085,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 4454
    },
    {
      "epoch": 1.7798641630043948,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 4455
    },
    {
      "epoch": 1.7802636835797043,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 4456
    },
    {
      "epoch": 1.780663204155014,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 4457
    },
    {
      "epoch": 1.7810627247303237,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 4458
    },
    {
      "epoch": 1.7814622453056332,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 4459
    },
    {
      "epoch": 1.7818617658809428,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 4460
    },
    {
      "epoch": 1.7822612864562526,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 4461
    },
    {
      "epoch": 1.7826608070315622,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 4462
    },
    {
      "epoch": 1.7830603276068717,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 4463
    },
    {
      "epoch": 1.7834598481821815,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0826,
      "step": 4464
    },
    {
      "epoch": 1.7838593687574908,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 4465
    },
    {
      "epoch": 1.7842588893328006,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 4466
    },
    {
      "epoch": 1.7846584099081104,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0774,
      "step": 4467
    },
    {
      "epoch": 1.7850579304834198,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 4468
    },
    {
      "epoch": 1.7854574510587295,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.075,
      "step": 4469
    },
    {
      "epoch": 1.7858569716340391,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.07,
      "step": 4470
    },
    {
      "epoch": 1.7862564922093487,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0679,
      "step": 4471
    },
    {
      "epoch": 1.7866560127846585,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 4472
    },
    {
      "epoch": 1.787055533359968,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 4473
    },
    {
      "epoch": 1.7874550539352776,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 4474
    },
    {
      "epoch": 1.7878545745105874,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 4475
    },
    {
      "epoch": 1.788254095085897,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0743,
      "step": 4476
    },
    {
      "epoch": 1.7886536156612065,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 4477
    },
    {
      "epoch": 1.7890531362365163,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 4478
    },
    {
      "epoch": 1.7894526568118259,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 4479
    },
    {
      "epoch": 1.7898521773871354,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 4480
    },
    {
      "epoch": 1.7902516979624452,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 4481
    },
    {
      "epoch": 1.7906512185377546,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0802,
      "step": 4482
    },
    {
      "epoch": 1.7910507391130643,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 4483
    },
    {
      "epoch": 1.791450259688374,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 4484
    },
    {
      "epoch": 1.7918497802636835,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0741,
      "step": 4485
    },
    {
      "epoch": 1.7922493008389933,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 4486
    },
    {
      "epoch": 1.7926488214143028,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 4487
    },
    {
      "epoch": 1.7930483419896124,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 4488
    },
    {
      "epoch": 1.7934478625649222,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0696,
      "step": 4489
    },
    {
      "epoch": 1.7938473831402317,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 4490
    },
    {
      "epoch": 1.7942469037155413,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 4491
    },
    {
      "epoch": 1.794646424290851,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 4492
    },
    {
      "epoch": 1.7950459448661606,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0823,
      "step": 4493
    },
    {
      "epoch": 1.7954454654414702,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 4494
    },
    {
      "epoch": 1.79584498601678,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 4495
    },
    {
      "epoch": 1.7962445065920893,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 4496
    },
    {
      "epoch": 1.7966440271673991,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 4497
    },
    {
      "epoch": 1.797043547742709,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0728,
      "step": 4498
    },
    {
      "epoch": 1.7974430683180183,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 4499
    },
    {
      "epoch": 1.797842588893328,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0742,
      "step": 4500
    },
    {
      "epoch": 1.7982421094686376,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 4501
    },
    {
      "epoch": 1.7986416300439472,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 4502
    },
    {
      "epoch": 1.799041150619257,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 4503
    },
    {
      "epoch": 1.7994406711945665,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 4504
    },
    {
      "epoch": 1.799840191769876,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 4505
    },
    {
      "epoch": 1.8002397123451859,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0737,
      "step": 4506
    },
    {
      "epoch": 1.8006392329204954,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 4507
    },
    {
      "epoch": 1.801038753495805,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 4508
    },
    {
      "epoch": 1.8014382740711148,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0756,
      "step": 4509
    },
    {
      "epoch": 1.8018377946464241,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 4510
    },
    {
      "epoch": 1.802237315221734,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 4511
    },
    {
      "epoch": 1.8026368357970437,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 4512
    },
    {
      "epoch": 1.803036356372353,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 4513
    },
    {
      "epoch": 1.8034358769476628,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0657,
      "step": 4514
    },
    {
      "epoch": 1.8038353975229724,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 4515
    },
    {
      "epoch": 1.804234918098282,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0779,
      "step": 4516
    },
    {
      "epoch": 1.8046344386735917,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0713,
      "step": 4517
    },
    {
      "epoch": 1.8050339592489013,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 4518
    },
    {
      "epoch": 1.8054334798242109,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 4519
    },
    {
      "epoch": 1.8058330003995207,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 4520
    },
    {
      "epoch": 1.8062325209748302,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 4521
    },
    {
      "epoch": 1.8066320415501398,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 4522
    },
    {
      "epoch": 1.8070315621254496,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 4523
    },
    {
      "epoch": 1.8074310827007591,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 4524
    },
    {
      "epoch": 1.8078306032760687,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0737,
      "step": 4525
    },
    {
      "epoch": 1.8082301238513785,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 4526
    },
    {
      "epoch": 1.8086296444266878,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 4527
    },
    {
      "epoch": 1.8090291650019976,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 4528
    },
    {
      "epoch": 1.8094286855773072,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 4529
    },
    {
      "epoch": 1.8098282061526167,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 4530
    },
    {
      "epoch": 1.8102277267279265,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 4531
    },
    {
      "epoch": 1.810627247303236,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 4532
    },
    {
      "epoch": 1.8110267678785457,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 4533
    },
    {
      "epoch": 1.8114262884538554,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 4534
    },
    {
      "epoch": 1.811825809029165,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 4535
    },
    {
      "epoch": 1.8122253296044746,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 4536
    },
    {
      "epoch": 1.8126248501797844,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 4537
    },
    {
      "epoch": 1.813024370755094,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0713,
      "step": 4538
    },
    {
      "epoch": 1.8134238913304035,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 4539
    },
    {
      "epoch": 1.8138234119057133,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 4540
    },
    {
      "epoch": 1.8142229324810226,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.082,
      "step": 4541
    },
    {
      "epoch": 1.8146224530563324,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 4542
    },
    {
      "epoch": 1.8150219736316422,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 4543
    },
    {
      "epoch": 1.8154214942069515,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 4544
    },
    {
      "epoch": 1.8158210147822613,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.075,
      "step": 4545
    },
    {
      "epoch": 1.8162205353575709,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0778,
      "step": 4546
    },
    {
      "epoch": 1.8166200559328805,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 4547
    },
    {
      "epoch": 1.8170195765081902,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 4548
    },
    {
      "epoch": 1.8174190970834998,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 4549
    },
    {
      "epoch": 1.8178186176588094,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0672,
      "step": 4550
    },
    {
      "epoch": 1.8182181382341192,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 4551
    },
    {
      "epoch": 1.8186176588094287,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0697,
      "step": 4552
    },
    {
      "epoch": 1.8190171793847383,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 4553
    },
    {
      "epoch": 1.819416699960048,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 4554
    },
    {
      "epoch": 1.8198162205353574,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 4555
    },
    {
      "epoch": 1.8202157411106672,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 4556
    },
    {
      "epoch": 1.820615261685977,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0736,
      "step": 4557
    },
    {
      "epoch": 1.8210147822612863,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 4558
    },
    {
      "epoch": 1.821414302836596,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 4559
    },
    {
      "epoch": 1.8218138234119057,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 4560
    },
    {
      "epoch": 1.8222133439872152,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 4561
    },
    {
      "epoch": 1.822612864562525,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 4562
    },
    {
      "epoch": 1.8230123851378346,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 4563
    },
    {
      "epoch": 1.8234119057131442,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 4564
    },
    {
      "epoch": 1.823811426288454,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 4565
    },
    {
      "epoch": 1.8242109468637635,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 4566
    },
    {
      "epoch": 1.824610467439073,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 4567
    },
    {
      "epoch": 1.8250099880143829,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 4568
    },
    {
      "epoch": 1.8254095085896924,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 4569
    },
    {
      "epoch": 1.825809029165002,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 4570
    },
    {
      "epoch": 1.8262085497403118,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.076,
      "step": 4571
    },
    {
      "epoch": 1.8266080703156211,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0672,
      "step": 4572
    },
    {
      "epoch": 1.827007590890931,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 4573
    },
    {
      "epoch": 1.8274071114662405,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 4574
    },
    {
      "epoch": 1.82780663204155,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 4575
    },
    {
      "epoch": 1.8282061526168598,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 4576
    },
    {
      "epoch": 1.8286056731921694,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0785,
      "step": 4577
    },
    {
      "epoch": 1.829005193767479,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 4578
    },
    {
      "epoch": 1.8294047143427887,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 4579
    },
    {
      "epoch": 1.8298042349180983,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0757,
      "step": 4580
    },
    {
      "epoch": 1.8302037554934079,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 4581
    },
    {
      "epoch": 1.8306032760687176,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0693,
      "step": 4582
    },
    {
      "epoch": 1.8310027966440272,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 4583
    },
    {
      "epoch": 1.8314023172193368,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 4584
    },
    {
      "epoch": 1.8318018377946466,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 4585
    },
    {
      "epoch": 1.832201358369956,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0706,
      "step": 4586
    },
    {
      "epoch": 1.8326008789452657,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 4587
    },
    {
      "epoch": 1.8330003995205755,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 4588
    },
    {
      "epoch": 1.8333999200958848,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 4589
    },
    {
      "epoch": 1.8337994406711946,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 4590
    },
    {
      "epoch": 1.8341989612465042,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0725,
      "step": 4591
    },
    {
      "epoch": 1.8345984818218137,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0706,
      "step": 4592
    },
    {
      "epoch": 1.8349980023971235,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.0728,
      "step": 4593
    },
    {
      "epoch": 1.835397522972433,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 4594
    },
    {
      "epoch": 1.8357970435477426,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 4595
    },
    {
      "epoch": 1.8361965641230524,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 4596
    },
    {
      "epoch": 1.836596084698362,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 4597
    },
    {
      "epoch": 1.8369956052736716,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 4598
    },
    {
      "epoch": 1.8373951258489813,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 4599
    },
    {
      "epoch": 1.8377946464242907,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 4600
    },
    {
      "epoch": 1.8381941669996005,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 4601
    },
    {
      "epoch": 1.8385936875749103,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 4602
    },
    {
      "epoch": 1.8389932081502196,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 4603
    },
    {
      "epoch": 1.8393927287255294,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 4604
    },
    {
      "epoch": 1.839792249300839,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 4605
    },
    {
      "epoch": 1.8401917698761485,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 4606
    },
    {
      "epoch": 1.8405912904514583,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 4607
    },
    {
      "epoch": 1.8409908110267679,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 4608
    },
    {
      "epoch": 1.8413903316020774,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 4609
    },
    {
      "epoch": 1.8417898521773872,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 4610
    },
    {
      "epoch": 1.8421893727526968,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 4611
    },
    {
      "epoch": 1.8425888933280063,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 4612
    },
    {
      "epoch": 1.8429884139033161,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 4613
    },
    {
      "epoch": 1.8433879344786257,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 4614
    },
    {
      "epoch": 1.8437874550539353,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 4615
    },
    {
      "epoch": 1.844186975629245,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 4616
    },
    {
      "epoch": 1.8445864962045544,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 4617
    },
    {
      "epoch": 1.8449860167798642,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 4618
    },
    {
      "epoch": 1.8453855373551737,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 4619
    },
    {
      "epoch": 1.8457850579304833,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 4620
    },
    {
      "epoch": 1.846184578505793,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0735,
      "step": 4621
    },
    {
      "epoch": 1.8465840990811027,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0665,
      "step": 4622
    },
    {
      "epoch": 1.8469836196564122,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 4623
    },
    {
      "epoch": 1.847383140231722,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 4624
    },
    {
      "epoch": 1.8477826608070316,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 4625
    },
    {
      "epoch": 1.8481821813823411,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 4626
    },
    {
      "epoch": 1.848581701957651,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 4627
    },
    {
      "epoch": 1.8489812225329605,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 4628
    },
    {
      "epoch": 1.84938074310827,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0792,
      "step": 4629
    },
    {
      "epoch": 1.8497802636835798,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 4630
    },
    {
      "epoch": 1.8501797842588892,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 4631
    },
    {
      "epoch": 1.850579304834199,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 4632
    },
    {
      "epoch": 1.8509788254095088,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0817,
      "step": 4633
    },
    {
      "epoch": 1.851378345984818,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 4634
    },
    {
      "epoch": 1.8517778665601279,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0817,
      "step": 4635
    },
    {
      "epoch": 1.8521773871354374,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 4636
    },
    {
      "epoch": 1.852576907710747,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 4637
    },
    {
      "epoch": 1.8529764282860568,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 4638
    },
    {
      "epoch": 1.8533759488613664,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 4639
    },
    {
      "epoch": 1.853775469436676,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 4640
    },
    {
      "epoch": 1.8541749900119857,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 4641
    },
    {
      "epoch": 1.8545745105872953,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 4642
    },
    {
      "epoch": 1.8549740311626048,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 4643
    },
    {
      "epoch": 1.8553735517379146,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 4644
    },
    {
      "epoch": 1.855773072313224,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.0598,
      "step": 4645
    },
    {
      "epoch": 1.8561725928885338,
      "grad_norm": 0.265625,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 4646
    },
    {
      "epoch": 1.8565721134638435,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 4647
    },
    {
      "epoch": 1.8569716340391529,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 4648
    },
    {
      "epoch": 1.8573711546144627,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 4649
    },
    {
      "epoch": 1.8577706751897722,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 4650
    },
    {
      "epoch": 1.8581701957650818,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 4651
    },
    {
      "epoch": 1.8585697163403916,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 4652
    },
    {
      "epoch": 1.8589692369157012,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 4653
    },
    {
      "epoch": 1.8593687574910107,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 4654
    },
    {
      "epoch": 1.8597682780663205,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 4655
    },
    {
      "epoch": 1.86016779864163,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 4656
    },
    {
      "epoch": 1.8605673192169396,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.0738,
      "step": 4657
    },
    {
      "epoch": 1.8609668397922494,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 4658
    },
    {
      "epoch": 1.861366360367559,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0738,
      "step": 4659
    },
    {
      "epoch": 1.8617658809428685,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 4660
    },
    {
      "epoch": 1.8621654015181783,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 4661
    },
    {
      "epoch": 1.8625649220934877,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 4662
    },
    {
      "epoch": 1.8629644426687975,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0716,
      "step": 4663
    },
    {
      "epoch": 1.863363963244107,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 4664
    },
    {
      "epoch": 1.8637634838194166,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0704,
      "step": 4665
    },
    {
      "epoch": 1.8641630043947264,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0857,
      "step": 4666
    },
    {
      "epoch": 1.864562524970036,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 4667
    },
    {
      "epoch": 1.8649620455453455,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 4668
    },
    {
      "epoch": 1.8653615661206553,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 4669
    },
    {
      "epoch": 1.8657610866959649,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 4670
    },
    {
      "epoch": 1.8661606072712744,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 4671
    },
    {
      "epoch": 1.8665601278465842,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 4672
    },
    {
      "epoch": 1.8669596484218938,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 4673
    },
    {
      "epoch": 1.8673591689972033,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 4674
    },
    {
      "epoch": 1.8677586895725131,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 4675
    },
    {
      "epoch": 1.8681582101478225,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 4676
    },
    {
      "epoch": 1.8685577307231322,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 4677
    },
    {
      "epoch": 1.868957251298442,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 4678
    },
    {
      "epoch": 1.8693567718737514,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 4679
    },
    {
      "epoch": 1.8697562924490612,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 4680
    },
    {
      "epoch": 1.8701558130243707,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 4681
    },
    {
      "epoch": 1.8705553335996803,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 4682
    },
    {
      "epoch": 1.87095485417499,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 4683
    },
    {
      "epoch": 1.8713543747502996,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 4684
    },
    {
      "epoch": 1.8717538953256092,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 4685
    },
    {
      "epoch": 1.872153415900919,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 4686
    },
    {
      "epoch": 1.8725529364762286,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 4687
    },
    {
      "epoch": 1.8729524570515381,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0734,
      "step": 4688
    },
    {
      "epoch": 1.873351977626848,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 4689
    },
    {
      "epoch": 1.8737514982021573,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 4690
    },
    {
      "epoch": 1.874151018777467,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 4691
    },
    {
      "epoch": 1.8745505393527768,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.074,
      "step": 4692
    },
    {
      "epoch": 1.8749500599280862,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.069,
      "step": 4693
    },
    {
      "epoch": 1.875349580503396,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 4694
    },
    {
      "epoch": 1.8757491010787055,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 4695
    },
    {
      "epoch": 1.876148621654015,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 4696
    },
    {
      "epoch": 1.8765481422293249,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0598,
      "step": 4697
    },
    {
      "epoch": 1.8769476628046344,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 4698
    },
    {
      "epoch": 1.877347183379944,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 4699
    },
    {
      "epoch": 1.8777467039552538,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 4700
    },
    {
      "epoch": 1.8781462245305633,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 4701
    },
    {
      "epoch": 1.878545745105873,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 4702
    },
    {
      "epoch": 1.8789452656811827,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 4703
    },
    {
      "epoch": 1.8793447862564923,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.081,
      "step": 4704
    },
    {
      "epoch": 1.8797443068318018,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 4705
    },
    {
      "epoch": 1.8801438274071116,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 4706
    },
    {
      "epoch": 1.880543347982421,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0657,
      "step": 4707
    },
    {
      "epoch": 1.8809428685577307,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 4708
    },
    {
      "epoch": 1.8813423891330403,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 4709
    },
    {
      "epoch": 1.8817419097083499,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 4710
    },
    {
      "epoch": 1.8821414302836597,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 4711
    },
    {
      "epoch": 1.8825409508589692,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 4712
    },
    {
      "epoch": 1.8829404714342788,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0741,
      "step": 4713
    },
    {
      "epoch": 1.8833399920095886,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 4714
    },
    {
      "epoch": 1.8837395125848981,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 4715
    },
    {
      "epoch": 1.8841390331602077,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0679,
      "step": 4716
    },
    {
      "epoch": 1.8845385537355175,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0764,
      "step": 4717
    },
    {
      "epoch": 1.884938074310827,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 4718
    },
    {
      "epoch": 1.8853375948861366,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.069,
      "step": 4719
    },
    {
      "epoch": 1.8857371154614464,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 4720
    },
    {
      "epoch": 1.8861366360367557,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 4721
    },
    {
      "epoch": 1.8865361566120655,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 4722
    },
    {
      "epoch": 1.8869356771873753,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 4723
    },
    {
      "epoch": 1.8873351977626847,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 4724
    },
    {
      "epoch": 1.8877347183379944,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 4725
    },
    {
      "epoch": 1.888134238913304,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 4726
    },
    {
      "epoch": 1.8885337594886136,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 4727
    },
    {
      "epoch": 1.8889332800639234,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 4728
    },
    {
      "epoch": 1.889332800639233,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 4729
    },
    {
      "epoch": 1.8897323212145425,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 4730
    },
    {
      "epoch": 1.8901318417898523,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 4731
    },
    {
      "epoch": 1.8905313623651618,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 4732
    },
    {
      "epoch": 1.8909308829404714,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 4733
    },
    {
      "epoch": 1.8913304035157812,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 4734
    },
    {
      "epoch": 1.8917299240910905,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 4735
    },
    {
      "epoch": 1.8921294446664003,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 4736
    },
    {
      "epoch": 1.89252896524171,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 4737
    },
    {
      "epoch": 1.8929284858170194,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0773,
      "step": 4738
    },
    {
      "epoch": 1.8933280063923292,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 4739
    },
    {
      "epoch": 1.8937275269676388,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 4740
    },
    {
      "epoch": 1.8941270475429484,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 4741
    },
    {
      "epoch": 1.8945265681182581,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 4742
    },
    {
      "epoch": 1.8949260886935677,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 4743
    },
    {
      "epoch": 1.8953256092688773,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 4744
    },
    {
      "epoch": 1.895725129844187,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 4745
    },
    {
      "epoch": 1.8961246504194966,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 4746
    },
    {
      "epoch": 1.8965241709948062,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 4747
    },
    {
      "epoch": 1.896923691570116,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 4748
    },
    {
      "epoch": 1.8973232121454255,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0725,
      "step": 4749
    },
    {
      "epoch": 1.897722732720735,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 4750
    },
    {
      "epoch": 1.898122253296045,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 4751
    },
    {
      "epoch": 1.8985217738713542,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 4752
    },
    {
      "epoch": 1.898921294446664,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 4753
    },
    {
      "epoch": 1.8993208150219736,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0721,
      "step": 4754
    },
    {
      "epoch": 1.8997203355972831,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 4755
    },
    {
      "epoch": 1.900119856172593,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 4756
    },
    {
      "epoch": 1.9005193767479025,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 4757
    },
    {
      "epoch": 1.900918897323212,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 4758
    },
    {
      "epoch": 1.9013184178985219,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 4759
    },
    {
      "epoch": 1.9017179384738314,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 4760
    },
    {
      "epoch": 1.902117459049141,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 4761
    },
    {
      "epoch": 1.9025169796244508,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 4762
    },
    {
      "epoch": 1.9029165001997603,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 4763
    },
    {
      "epoch": 1.90331602077507,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 4764
    },
    {
      "epoch": 1.9037155413503797,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 4765
    },
    {
      "epoch": 1.904115061925689,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 4766
    },
    {
      "epoch": 1.9045145825009988,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 4767
    },
    {
      "epoch": 1.9049141030763086,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 4768
    },
    {
      "epoch": 1.905313623651618,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 4769
    },
    {
      "epoch": 1.9057131442269277,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 4770
    },
    {
      "epoch": 1.9061126648022373,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 4771
    },
    {
      "epoch": 1.9065121853775469,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 4772
    },
    {
      "epoch": 1.9069117059528566,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 4773
    },
    {
      "epoch": 1.9073112265281662,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 4774
    },
    {
      "epoch": 1.9077107471034758,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 4775
    },
    {
      "epoch": 1.9081102676787856,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0737,
      "step": 4776
    },
    {
      "epoch": 1.9085097882540951,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 4777
    },
    {
      "epoch": 1.9089093088294047,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 4778
    },
    {
      "epoch": 1.9093088294047145,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 4779
    },
    {
      "epoch": 1.9097083499800238,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0657,
      "step": 4780
    },
    {
      "epoch": 1.9101078705553336,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0727,
      "step": 4781
    },
    {
      "epoch": 1.9105073911306434,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 4782
    },
    {
      "epoch": 1.9109069117059527,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 4783
    },
    {
      "epoch": 1.9113064322812625,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0766,
      "step": 4784
    },
    {
      "epoch": 1.911705952856572,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0672,
      "step": 4785
    },
    {
      "epoch": 1.9121054734318816,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 4786
    },
    {
      "epoch": 1.9125049940071914,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.074,
      "step": 4787
    },
    {
      "epoch": 1.912904514582501,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 4788
    },
    {
      "epoch": 1.9133040351578106,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 4789
    },
    {
      "epoch": 1.9137035557331203,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 4790
    },
    {
      "epoch": 1.91410307630843,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 4791
    },
    {
      "epoch": 1.9145025968837395,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0712,
      "step": 4792
    },
    {
      "epoch": 1.9149021174590493,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 4793
    },
    {
      "epoch": 1.9153016380343588,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 4794
    },
    {
      "epoch": 1.9157011586096684,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 4795
    },
    {
      "epoch": 1.9161006791849782,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0778,
      "step": 4796
    },
    {
      "epoch": 1.9165001997602875,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 4797
    },
    {
      "epoch": 1.9168997203355973,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 4798
    },
    {
      "epoch": 1.9172992409109069,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 4799
    },
    {
      "epoch": 1.9176987614862164,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 4800
    },
    {
      "epoch": 1.9180982820615262,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 4801
    },
    {
      "epoch": 1.9184978026368358,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 4802
    },
    {
      "epoch": 1.9188973232121453,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 4803
    },
    {
      "epoch": 1.9192968437874551,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 4804
    },
    {
      "epoch": 1.9196963643627647,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 4805
    },
    {
      "epoch": 1.9200958849380743,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0737,
      "step": 4806
    },
    {
      "epoch": 1.920495405513384,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 4807
    },
    {
      "epoch": 1.9208949260886936,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 4808
    },
    {
      "epoch": 1.9212944466640032,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.076,
      "step": 4809
    },
    {
      "epoch": 1.921693967239313,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 4810
    },
    {
      "epoch": 1.9220934878146223,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 4811
    },
    {
      "epoch": 1.922493008389932,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0713,
      "step": 4812
    },
    {
      "epoch": 1.9228925289652419,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 4813
    },
    {
      "epoch": 1.9232920495405512,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 4814
    },
    {
      "epoch": 1.923691570115861,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0805,
      "step": 4815
    },
    {
      "epoch": 1.9240910906911706,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 4816
    },
    {
      "epoch": 1.9244906112664801,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 4817
    },
    {
      "epoch": 1.92489013184179,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 4818
    },
    {
      "epoch": 1.9252896524170995,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 4819
    },
    {
      "epoch": 1.925689172992409,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 4820
    },
    {
      "epoch": 1.9260886935677188,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 4821
    },
    {
      "epoch": 1.9264882141430284,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 4822
    },
    {
      "epoch": 1.926887734718338,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0753,
      "step": 4823
    },
    {
      "epoch": 1.9272872552936477,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 4824
    },
    {
      "epoch": 1.927686775868957,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 4825
    },
    {
      "epoch": 1.9280862964442669,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 4826
    },
    {
      "epoch": 1.9284858170195767,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 4827
    },
    {
      "epoch": 1.928885337594886,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 4828
    },
    {
      "epoch": 1.9292848581701958,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 4829
    },
    {
      "epoch": 1.9296843787455054,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 4830
    },
    {
      "epoch": 1.930083899320815,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 4831
    },
    {
      "epoch": 1.9304834198961247,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 4832
    },
    {
      "epoch": 1.9308829404714343,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 4833
    },
    {
      "epoch": 1.9312824610467438,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 4834
    },
    {
      "epoch": 1.9316819816220536,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 4835
    },
    {
      "epoch": 1.9320815021973632,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 4836
    },
    {
      "epoch": 1.9324810227726728,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 4837
    },
    {
      "epoch": 1.9328805433479825,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 4838
    },
    {
      "epoch": 1.933280063923292,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 4839
    },
    {
      "epoch": 1.9336795844986017,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 4840
    },
    {
      "epoch": 1.9340791050739115,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0766,
      "step": 4841
    },
    {
      "epoch": 1.9344786256492208,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 4842
    },
    {
      "epoch": 1.9348781462245306,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 4843
    },
    {
      "epoch": 1.9352776667998401,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 4844
    },
    {
      "epoch": 1.9356771873751497,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 4845
    },
    {
      "epoch": 1.9360767079504595,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0665,
      "step": 4846
    },
    {
      "epoch": 1.936476228525769,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 4847
    },
    {
      "epoch": 1.9368757491010786,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0753,
      "step": 4848
    },
    {
      "epoch": 1.9372752696763884,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 4849
    },
    {
      "epoch": 1.937674790251698,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 4850
    },
    {
      "epoch": 1.9380743108270075,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 4851
    },
    {
      "epoch": 1.9384738314023173,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 4852
    },
    {
      "epoch": 1.938873351977627,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 4853
    },
    {
      "epoch": 1.9392728725529365,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 4854
    },
    {
      "epoch": 1.9396723931282462,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 4855
    },
    {
      "epoch": 1.9400719137035556,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 4856
    },
    {
      "epoch": 1.9404714342788654,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 4857
    },
    {
      "epoch": 1.940870954854175,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0721,
      "step": 4858
    },
    {
      "epoch": 1.9412704754294845,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 4859
    },
    {
      "epoch": 1.9416699960047943,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0713,
      "step": 4860
    },
    {
      "epoch": 1.9420695165801038,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 4861
    },
    {
      "epoch": 1.9424690371554134,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 4862
    },
    {
      "epoch": 1.9428685577307232,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 4863
    },
    {
      "epoch": 1.9432680783060328,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 4864
    },
    {
      "epoch": 1.9436675988813423,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 4865
    },
    {
      "epoch": 1.9440671194566521,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 4866
    },
    {
      "epoch": 1.9444666400319617,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 4867
    },
    {
      "epoch": 1.9448661606072712,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0772,
      "step": 4868
    },
    {
      "epoch": 1.945265681182581,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 4869
    },
    {
      "epoch": 1.9456652017578904,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 4870
    },
    {
      "epoch": 1.9460647223332002,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 4871
    },
    {
      "epoch": 1.94646424290851,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 4872
    },
    {
      "epoch": 1.9468637634838193,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 4873
    },
    {
      "epoch": 1.947263284059129,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 4874
    },
    {
      "epoch": 1.9476628046344386,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 4875
    },
    {
      "epoch": 1.9480623252097482,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0736,
      "step": 4876
    },
    {
      "epoch": 1.948461845785058,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.078,
      "step": 4877
    },
    {
      "epoch": 1.9488613663603676,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 4878
    },
    {
      "epoch": 1.9492608869356771,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 4879
    },
    {
      "epoch": 1.949660407510987,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 4880
    },
    {
      "epoch": 1.9500599280862965,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 4881
    },
    {
      "epoch": 1.950459448661606,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 4882
    },
    {
      "epoch": 1.9508589692369158,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 4883
    },
    {
      "epoch": 1.9512584898122254,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 4884
    },
    {
      "epoch": 1.951658010387535,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 4885
    },
    {
      "epoch": 1.9520575309628447,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 4886
    },
    {
      "epoch": 1.952457051538154,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 4887
    },
    {
      "epoch": 1.9528565721134639,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 4888
    },
    {
      "epoch": 1.9532560926887734,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 4889
    },
    {
      "epoch": 1.953655613264083,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 4890
    },
    {
      "epoch": 1.9540551338393928,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 4891
    },
    {
      "epoch": 1.9544546544147023,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 4892
    },
    {
      "epoch": 1.954854174990012,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 4893
    },
    {
      "epoch": 1.9552536955653217,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0743,
      "step": 4894
    },
    {
      "epoch": 1.9556532161406313,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 4895
    },
    {
      "epoch": 1.9560527367159408,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 4896
    },
    {
      "epoch": 1.9564522572912506,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 4897
    },
    {
      "epoch": 1.9568517778665602,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 4898
    },
    {
      "epoch": 1.9572512984418697,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 4899
    },
    {
      "epoch": 1.9576508190171795,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 4900
    },
    {
      "epoch": 1.9580503395924889,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 4901
    },
    {
      "epoch": 1.9584498601677987,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 4902
    },
    {
      "epoch": 1.9588493807431082,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 4903
    },
    {
      "epoch": 1.9592489013184178,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0706,
      "step": 4904
    },
    {
      "epoch": 1.9596484218937276,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 4905
    },
    {
      "epoch": 1.9600479424690371,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 4906
    },
    {
      "epoch": 1.9604474630443467,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 4907
    },
    {
      "epoch": 1.9608469836196565,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 4908
    },
    {
      "epoch": 1.961246504194966,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 4909
    },
    {
      "epoch": 1.9616460247702756,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 4910
    },
    {
      "epoch": 1.9620455453455854,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 4911
    },
    {
      "epoch": 1.962445065920895,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 4912
    },
    {
      "epoch": 1.9628445864962045,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 4913
    },
    {
      "epoch": 1.9632441070715143,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 4914
    },
    {
      "epoch": 1.9636436276468237,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 4915
    },
    {
      "epoch": 1.9640431482221334,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.074,
      "step": 4916
    },
    {
      "epoch": 1.9644426687974432,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 4917
    },
    {
      "epoch": 1.9648421893727526,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 4918
    },
    {
      "epoch": 1.9652417099480624,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 4919
    },
    {
      "epoch": 1.965641230523372,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 4920
    },
    {
      "epoch": 1.9660407510986815,
      "grad_norm": 0.375,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 4921
    },
    {
      "epoch": 1.9664402716739913,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 4922
    },
    {
      "epoch": 1.9668397922493008,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 4923
    },
    {
      "epoch": 1.9672393128246104,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005,
      "loss": 1.0715,
      "step": 4924
    },
    {
      "epoch": 1.9676388333999202,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 4925
    },
    {
      "epoch": 1.9680383539752297,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 4926
    },
    {
      "epoch": 1.9684378745505393,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 4927
    },
    {
      "epoch": 1.968837395125849,
      "grad_norm": 0.375,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 4928
    },
    {
      "epoch": 1.9692369157011587,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 4929
    },
    {
      "epoch": 1.9696364362764682,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 4930
    },
    {
      "epoch": 1.970035956851778,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 4931
    },
    {
      "epoch": 1.9704354774270874,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 4932
    },
    {
      "epoch": 1.9708349980023971,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 4933
    },
    {
      "epoch": 1.9712345185777067,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 4934
    },
    {
      "epoch": 1.9716340391530163,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 4935
    },
    {
      "epoch": 1.972033559728326,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0005,
      "loss": 1.0756,
      "step": 4936
    },
    {
      "epoch": 1.9724330803036356,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0731,
      "step": 4937
    },
    {
      "epoch": 1.9728326008789452,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 4938
    },
    {
      "epoch": 1.973232121454255,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 4939
    },
    {
      "epoch": 1.9736316420295645,
      "grad_norm": 0.546875,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 4940
    },
    {
      "epoch": 1.974031162604874,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 4941
    },
    {
      "epoch": 1.9744306831801839,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 4942
    },
    {
      "epoch": 1.9748302037554935,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 4943
    },
    {
      "epoch": 1.975229724330803,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 4944
    },
    {
      "epoch": 1.9756292449061128,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 4945
    },
    {
      "epoch": 1.9760287654814221,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 4946
    },
    {
      "epoch": 1.976428286056732,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0005,
      "loss": 1.0741,
      "step": 4947
    },
    {
      "epoch": 1.9768278066320415,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 4948
    },
    {
      "epoch": 1.977227327207351,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 4949
    },
    {
      "epoch": 1.9776268477826608,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 4950
    },
    {
      "epoch": 1.9780263683579704,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 4951
    },
    {
      "epoch": 1.97842588893328,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 4952
    },
    {
      "epoch": 1.9788254095085898,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0005,
      "loss": 1.0777,
      "step": 4953
    },
    {
      "epoch": 1.9792249300838993,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 4954
    },
    {
      "epoch": 1.979624450659209,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 4955
    },
    {
      "epoch": 1.9800239712345187,
      "grad_norm": 0.203125,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 4956
    },
    {
      "epoch": 1.9804234918098282,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 4957
    },
    {
      "epoch": 1.9808230123851378,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 4958
    },
    {
      "epoch": 1.9812225329604476,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 4959
    },
    {
      "epoch": 1.981622053535757,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 4960
    },
    {
      "epoch": 1.9820215741110667,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 4961
    },
    {
      "epoch": 1.9824210946863765,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 4962
    },
    {
      "epoch": 1.9828206152616858,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 4963
    },
    {
      "epoch": 1.9832201358369956,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 4964
    },
    {
      "epoch": 1.9836196564123052,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 4965
    },
    {
      "epoch": 1.9840191769876148,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 4966
    },
    {
      "epoch": 1.9844186975629245,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0722,
      "step": 4967
    },
    {
      "epoch": 1.9848182181382341,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 4968
    },
    {
      "epoch": 1.9852177387135437,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 4969
    },
    {
      "epoch": 1.9856172592888535,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 4970
    },
    {
      "epoch": 1.986016779864163,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 4971
    },
    {
      "epoch": 1.9864163004394726,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 4972
    },
    {
      "epoch": 1.9868158210147824,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 4973
    },
    {
      "epoch": 1.987215341590092,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 4974
    },
    {
      "epoch": 1.9876148621654015,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 4975
    },
    {
      "epoch": 1.9880143827407113,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 4976
    },
    {
      "epoch": 1.9884139033160206,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 4977
    },
    {
      "epoch": 1.9888134238913304,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 4978
    },
    {
      "epoch": 1.98921294446664,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0716,
      "step": 4979
    },
    {
      "epoch": 1.9896124650419496,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 4980
    },
    {
      "epoch": 1.9900119856172593,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 4981
    },
    {
      "epoch": 1.990411506192569,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 4982
    },
    {
      "epoch": 1.9908110267678785,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 4983
    },
    {
      "epoch": 1.9912105473431883,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 4984
    },
    {
      "epoch": 1.9916100679184978,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 4985
    },
    {
      "epoch": 1.9920095884938074,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 4986
    },
    {
      "epoch": 1.9924091090691172,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 4987
    },
    {
      "epoch": 1.9928086296444267,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 4988
    },
    {
      "epoch": 1.9932081502197363,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 4989
    },
    {
      "epoch": 1.993607670795046,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 4990
    },
    {
      "epoch": 1.9940071913703554,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 4991
    },
    {
      "epoch": 1.9944067119456652,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 4992
    },
    {
      "epoch": 1.9948062325209748,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 4993
    },
    {
      "epoch": 1.9952057530962843,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 4994
    },
    {
      "epoch": 1.9956052736715941,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 4995
    },
    {
      "epoch": 1.9960047942469037,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 4996
    },
    {
      "epoch": 1.9964043148222133,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 4997
    },
    {
      "epoch": 1.996803835397523,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0721,
      "step": 4998
    },
    {
      "epoch": 1.9972033559728326,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 4999
    },
    {
      "epoch": 1.9976028765481422,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 5000
    },
    {
      "epoch": 1.998002397123452,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 5001
    },
    {
      "epoch": 1.9984019176987615,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 5002
    },
    {
      "epoch": 1.998801438274071,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 5003
    },
    {
      "epoch": 1.9992009588493809,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 5004
    },
    {
      "epoch": 1.9996004794246902,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 5005
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 5006
    },
    {
      "epoch": 2.00039952057531,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 5007
    },
    {
      "epoch": 2.000799041150619,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 5008
    },
    {
      "epoch": 2.001198561725929,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0725,
      "step": 5009
    },
    {
      "epoch": 2.0015980823012387,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 5010
    },
    {
      "epoch": 2.001997602876548,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 5011
    },
    {
      "epoch": 2.002397123451858,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 5012
    },
    {
      "epoch": 2.0027966440271676,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 5013
    },
    {
      "epoch": 2.003196164602477,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0759,
      "step": 5014
    },
    {
      "epoch": 2.0035956851777867,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0711,
      "step": 5015
    },
    {
      "epoch": 2.003995205753096,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 5016
    },
    {
      "epoch": 2.004394726328406,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 5017
    },
    {
      "epoch": 2.0047942469037157,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 5018
    },
    {
      "epoch": 2.005193767479025,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 5019
    },
    {
      "epoch": 2.005593288054335,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 5020
    },
    {
      "epoch": 2.0059928086296446,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 5021
    },
    {
      "epoch": 2.006392329204954,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 5022
    },
    {
      "epoch": 2.0067918497802637,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 5023
    },
    {
      "epoch": 2.0071913703555735,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 5024
    },
    {
      "epoch": 2.007590890930883,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 5025
    },
    {
      "epoch": 2.0079904115061926,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0697,
      "step": 5026
    },
    {
      "epoch": 2.0083899320815024,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 5027
    },
    {
      "epoch": 2.0087894526568117,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 5028
    },
    {
      "epoch": 2.0091889732321215,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 5029
    },
    {
      "epoch": 2.009588493807431,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 5030
    },
    {
      "epoch": 2.0099880143827407,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 5031
    },
    {
      "epoch": 2.0103875349580504,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 5032
    },
    {
      "epoch": 2.01078705553336,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 5033
    },
    {
      "epoch": 2.0111865761086696,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 5034
    },
    {
      "epoch": 2.0115860966839794,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 5035
    },
    {
      "epoch": 2.0119856172592887,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 5036
    },
    {
      "epoch": 2.0123851378345985,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 5037
    },
    {
      "epoch": 2.0127846584099083,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 5038
    },
    {
      "epoch": 2.0131841789852176,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 5039
    },
    {
      "epoch": 2.0135836995605274,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 5040
    },
    {
      "epoch": 2.013983220135837,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 5041
    },
    {
      "epoch": 2.0143827407111465,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0732,
      "step": 5042
    },
    {
      "epoch": 2.0147822612864563,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.079,
      "step": 5043
    },
    {
      "epoch": 2.0151817818617657,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 5044
    },
    {
      "epoch": 2.0155813024370755,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 5045
    },
    {
      "epoch": 2.0159808230123852,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 5046
    },
    {
      "epoch": 2.0163803435876946,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 5047
    },
    {
      "epoch": 2.0167798641630044,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 5048
    },
    {
      "epoch": 2.017179384738314,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 5049
    },
    {
      "epoch": 2.0175789053136235,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 5050
    },
    {
      "epoch": 2.0179784258889333,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 5051
    },
    {
      "epoch": 2.018377946464243,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 5052
    },
    {
      "epoch": 2.0187774670395524,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 5053
    },
    {
      "epoch": 2.019176987614862,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 5054
    },
    {
      "epoch": 2.019576508190172,
      "grad_norm": 0.171875,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 5055
    },
    {
      "epoch": 2.0199760287654813,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 5056
    },
    {
      "epoch": 2.020375549340791,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 5057
    },
    {
      "epoch": 2.020775069916101,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 5058
    },
    {
      "epoch": 2.0211745904914102,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 5059
    },
    {
      "epoch": 2.02157411106672,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 5060
    },
    {
      "epoch": 2.0219736316420294,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 5061
    },
    {
      "epoch": 2.022373152217339,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 5062
    },
    {
      "epoch": 2.022772672792649,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 5063
    },
    {
      "epoch": 2.0231721933679583,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 5064
    },
    {
      "epoch": 2.023571713943268,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 5065
    },
    {
      "epoch": 2.023971234518578,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 5066
    },
    {
      "epoch": 2.024370755093887,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 5067
    },
    {
      "epoch": 2.024770275669197,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0651,
      "step": 5068
    },
    {
      "epoch": 2.0251697962445068,
      "grad_norm": 0.234375,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 5069
    },
    {
      "epoch": 2.025569316819816,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 5070
    },
    {
      "epoch": 2.025968837395126,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 5071
    },
    {
      "epoch": 2.0263683579704357,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0774,
      "step": 5072
    },
    {
      "epoch": 2.026767878545745,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 5073
    },
    {
      "epoch": 2.027167399121055,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 5074
    },
    {
      "epoch": 2.027566919696364,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 5075
    },
    {
      "epoch": 2.027966440271674,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 5076
    },
    {
      "epoch": 2.0283659608469837,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 5077
    },
    {
      "epoch": 2.028765481422293,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 5078
    },
    {
      "epoch": 2.029165001997603,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 5079
    },
    {
      "epoch": 2.0295645225729126,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 5080
    },
    {
      "epoch": 2.029964043148222,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 5081
    },
    {
      "epoch": 2.0303635637235318,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 5082
    },
    {
      "epoch": 2.0307630842988416,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 5083
    },
    {
      "epoch": 2.031162604874151,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 5084
    },
    {
      "epoch": 2.0315621254494607,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 5085
    },
    {
      "epoch": 2.0319616460247705,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 5086
    },
    {
      "epoch": 2.03236116660008,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 5087
    },
    {
      "epoch": 2.0327606871753896,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 5088
    },
    {
      "epoch": 2.033160207750699,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 5089
    },
    {
      "epoch": 2.0335597283260087,
      "grad_norm": 0.171875,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 5090
    },
    {
      "epoch": 2.0339592489013185,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 5091
    },
    {
      "epoch": 2.034358769476628,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 5092
    },
    {
      "epoch": 2.0347582900519376,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 5093
    },
    {
      "epoch": 2.0351578106272474,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 5094
    },
    {
      "epoch": 2.0355573312025568,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 5095
    },
    {
      "epoch": 2.0359568517778666,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 5096
    },
    {
      "epoch": 2.0363563723531763,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 5097
    },
    {
      "epoch": 2.0367558929284857,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0696,
      "step": 5098
    },
    {
      "epoch": 2.0371554135037955,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 5099
    },
    {
      "epoch": 2.0375549340791053,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0005,
      "loss": 1.0713,
      "step": 5100
    },
    {
      "epoch": 2.0379544546544146,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 5101
    },
    {
      "epoch": 2.0383539752297244,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 5102
    },
    {
      "epoch": 2.038753495805034,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 5103
    },
    {
      "epoch": 2.0391530163803435,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0005,
      "loss": 1.0598,
      "step": 5104
    },
    {
      "epoch": 2.0395525369556533,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 5105
    },
    {
      "epoch": 2.0399520575309626,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 5106
    },
    {
      "epoch": 2.0403515781062724,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0762,
      "step": 5107
    },
    {
      "epoch": 2.040751098681582,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0719,
      "step": 5108
    },
    {
      "epoch": 2.0411506192568916,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0766,
      "step": 5109
    },
    {
      "epoch": 2.0415501398322013,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 5110
    },
    {
      "epoch": 2.041949660407511,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0738,
      "step": 5111
    },
    {
      "epoch": 2.0423491809828205,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 5112
    },
    {
      "epoch": 2.0427487015581303,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 5113
    },
    {
      "epoch": 2.04314822213344,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 5114
    },
    {
      "epoch": 2.0435477427087494,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0727,
      "step": 5115
    },
    {
      "epoch": 2.043947263284059,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 5116
    },
    {
      "epoch": 2.044346783859369,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0751,
      "step": 5117
    },
    {
      "epoch": 2.0447463044346783,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 5118
    },
    {
      "epoch": 2.045145825009988,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 5119
    },
    {
      "epoch": 2.0455453455852974,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.0693,
      "step": 5120
    },
    {
      "epoch": 2.0459448661606072,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 5121
    },
    {
      "epoch": 2.046344386735917,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 5122
    },
    {
      "epoch": 2.0467439073112264,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 5123
    },
    {
      "epoch": 2.047143427886536,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 5124
    },
    {
      "epoch": 2.047542948461846,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 5125
    },
    {
      "epoch": 2.0479424690371553,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 5126
    },
    {
      "epoch": 2.048341989612465,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0687,
      "step": 5127
    },
    {
      "epoch": 2.048741510187775,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0732,
      "step": 5128
    },
    {
      "epoch": 2.049141030763084,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 5129
    },
    {
      "epoch": 2.049540551338394,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 5130
    },
    {
      "epoch": 2.0499400719137038,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 5131
    },
    {
      "epoch": 2.050339592489013,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0755,
      "step": 5132
    },
    {
      "epoch": 2.050739113064323,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 5133
    },
    {
      "epoch": 2.0511386336396322,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 5134
    },
    {
      "epoch": 2.051538154214942,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 5135
    },
    {
      "epoch": 2.051937674790252,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0687,
      "step": 5136
    },
    {
      "epoch": 2.052337195365561,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0657,
      "step": 5137
    },
    {
      "epoch": 2.052736715940871,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0782,
      "step": 5138
    },
    {
      "epoch": 2.0531362365161807,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 5139
    },
    {
      "epoch": 2.05353575709149,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 5140
    },
    {
      "epoch": 2.0539352776668,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 5141
    },
    {
      "epoch": 2.0543347982421096,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 5142
    },
    {
      "epoch": 2.054734318817419,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 5143
    },
    {
      "epoch": 2.0551338393927288,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 5144
    },
    {
      "epoch": 2.0555333599680385,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 5145
    },
    {
      "epoch": 2.055932880543348,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 5146
    },
    {
      "epoch": 2.0563324011186577,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 5147
    },
    {
      "epoch": 2.0567319216939675,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 5148
    },
    {
      "epoch": 2.057131442269277,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 5149
    },
    {
      "epoch": 2.0575309628445866,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 5150
    },
    {
      "epoch": 2.057930483419896,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 5151
    },
    {
      "epoch": 2.0583300039952057,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 5152
    },
    {
      "epoch": 2.0587295245705155,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 5153
    },
    {
      "epoch": 2.059129045145825,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 5154
    },
    {
      "epoch": 2.0595285657211346,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 5155
    },
    {
      "epoch": 2.0599280862964444,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 5156
    },
    {
      "epoch": 2.0603276068717538,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 5157
    },
    {
      "epoch": 2.0607271274470635,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 5158
    },
    {
      "epoch": 2.0611266480223733,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 5159
    },
    {
      "epoch": 2.0615261685976827,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 5160
    },
    {
      "epoch": 2.0619256891729925,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 5161
    },
    {
      "epoch": 2.0623252097483022,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 5162
    },
    {
      "epoch": 2.0627247303236116,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 5163
    },
    {
      "epoch": 2.0631242508989214,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0711,
      "step": 5164
    },
    {
      "epoch": 2.0635237714742307,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 5165
    },
    {
      "epoch": 2.0639232920495405,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 5166
    },
    {
      "epoch": 2.0643228126248503,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 5167
    },
    {
      "epoch": 2.0647223332001596,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 5168
    },
    {
      "epoch": 2.0651218537754694,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 5169
    },
    {
      "epoch": 2.065521374350779,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 5170
    },
    {
      "epoch": 2.0659208949260885,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 5171
    },
    {
      "epoch": 2.0663204155013983,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0679,
      "step": 5172
    },
    {
      "epoch": 2.066719936076708,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 5173
    },
    {
      "epoch": 2.0671194566520175,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 5174
    },
    {
      "epoch": 2.0675189772273272,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 5175
    },
    {
      "epoch": 2.067918497802637,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 5176
    },
    {
      "epoch": 2.0683180183779464,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 5177
    },
    {
      "epoch": 2.068717538953256,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 5178
    },
    {
      "epoch": 2.0691170595285655,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 5179
    },
    {
      "epoch": 2.0695165801038753,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0847,
      "step": 5180
    },
    {
      "epoch": 2.069916100679185,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 5181
    },
    {
      "epoch": 2.0703156212544944,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 5182
    },
    {
      "epoch": 2.070715141829804,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 5183
    },
    {
      "epoch": 2.071114662405114,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 5184
    },
    {
      "epoch": 2.0715141829804233,
      "grad_norm": 0.25,
      "learning_rate": 0.0005,
      "loss": 1.0721,
      "step": 5185
    },
    {
      "epoch": 2.071913703555733,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 5186
    },
    {
      "epoch": 2.072313224131043,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 5187
    },
    {
      "epoch": 2.0727127447063523,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 5188
    },
    {
      "epoch": 2.073112265281662,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 5189
    },
    {
      "epoch": 2.073511785856972,
      "grad_norm": 0.609375,
      "learning_rate": 0.0005,
      "loss": 1.0752,
      "step": 5190
    },
    {
      "epoch": 2.073911306432281,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0768,
      "step": 5191
    },
    {
      "epoch": 2.074310827007591,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 5192
    },
    {
      "epoch": 2.0747103475829007,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 5193
    },
    {
      "epoch": 2.07510986815821,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 5194
    },
    {
      "epoch": 2.07550938873352,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 5195
    },
    {
      "epoch": 2.075908909308829,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 5196
    },
    {
      "epoch": 2.076308429884139,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 5197
    },
    {
      "epoch": 2.076707950459449,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 5198
    },
    {
      "epoch": 2.077107471034758,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 5199
    },
    {
      "epoch": 2.077506991610068,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0766,
      "step": 5200
    },
    {
      "epoch": 2.0779065121853777,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 5201
    },
    {
      "epoch": 2.078306032760687,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 5202
    },
    {
      "epoch": 2.078705553335997,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 5203
    },
    {
      "epoch": 2.0791050739113066,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 5204
    },
    {
      "epoch": 2.079504594486616,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0812,
      "step": 5205
    },
    {
      "epoch": 2.0799041150619257,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 5206
    },
    {
      "epoch": 2.0803036356372355,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 5207
    },
    {
      "epoch": 2.080703156212545,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 5208
    },
    {
      "epoch": 2.0811026767878547,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0711,
      "step": 5209
    },
    {
      "epoch": 2.081502197363164,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 5210
    },
    {
      "epoch": 2.081901717938474,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 5211
    },
    {
      "epoch": 2.0823012385137836,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 5212
    },
    {
      "epoch": 2.082700759089093,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 5213
    },
    {
      "epoch": 2.0831002796644027,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 5214
    },
    {
      "epoch": 2.0834998002397125,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 5215
    },
    {
      "epoch": 2.083899320815022,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 5216
    },
    {
      "epoch": 2.0842988413903316,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0672,
      "step": 5217
    },
    {
      "epoch": 2.0846983619656414,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 5218
    },
    {
      "epoch": 2.0850978825409507,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 5219
    },
    {
      "epoch": 2.0854974031162605,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 5220
    },
    {
      "epoch": 2.0858969236915703,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 5221
    },
    {
      "epoch": 2.0862964442668797,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.078,
      "step": 5222
    },
    {
      "epoch": 2.0866959648421894,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 5223
    },
    {
      "epoch": 2.087095485417499,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 5224
    },
    {
      "epoch": 2.0874950059928086,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 5225
    },
    {
      "epoch": 2.0878945265681184,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0697,
      "step": 5226
    },
    {
      "epoch": 2.0882940471434277,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 5227
    },
    {
      "epoch": 2.0886935677187375,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0712,
      "step": 5228
    },
    {
      "epoch": 2.0890930882940473,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 5229
    },
    {
      "epoch": 2.0894926088693566,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 5230
    },
    {
      "epoch": 2.0898921294446664,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 5231
    },
    {
      "epoch": 2.090291650019976,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 5232
    },
    {
      "epoch": 2.0906911705952855,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 5233
    },
    {
      "epoch": 2.0910906911705953,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 5234
    },
    {
      "epoch": 2.091490211745905,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 5235
    },
    {
      "epoch": 2.0918897323212144,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 5236
    },
    {
      "epoch": 2.0922892528965242,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 5237
    },
    {
      "epoch": 2.092688773471834,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 5238
    },
    {
      "epoch": 2.0930882940471434,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 5239
    },
    {
      "epoch": 2.093487814622453,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 5240
    },
    {
      "epoch": 2.0938873351977625,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 5241
    },
    {
      "epoch": 2.0942868557730723,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 5242
    },
    {
      "epoch": 2.094686376348382,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 5243
    },
    {
      "epoch": 2.0950858969236914,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 5244
    },
    {
      "epoch": 2.095485417499001,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 5245
    },
    {
      "epoch": 2.095884938074311,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 5246
    },
    {
      "epoch": 2.0962844586496203,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 5247
    },
    {
      "epoch": 2.09668397922493,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0005,
      "loss": 1.0767,
      "step": 5248
    },
    {
      "epoch": 2.09708349980024,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 5249
    },
    {
      "epoch": 2.0974830203755492,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 5250
    },
    {
      "epoch": 2.097882540950859,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 5251
    },
    {
      "epoch": 2.098282061526169,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0743,
      "step": 5252
    },
    {
      "epoch": 2.098681582101478,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 5253
    },
    {
      "epoch": 2.099081102676788,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 5254
    },
    {
      "epoch": 2.0994806232520973,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 5255
    },
    {
      "epoch": 2.099880143827407,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 5256
    },
    {
      "epoch": 2.100279664402717,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 5257
    },
    {
      "epoch": 2.100679184978026,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 5258
    },
    {
      "epoch": 2.101078705553336,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 5259
    },
    {
      "epoch": 2.1014782261286458,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 5260
    },
    {
      "epoch": 2.101877746703955,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 5261
    },
    {
      "epoch": 2.102277267279265,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0771,
      "step": 5262
    },
    {
      "epoch": 2.1026767878545747,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 5263
    },
    {
      "epoch": 2.103076308429884,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 5264
    },
    {
      "epoch": 2.103475829005194,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 5265
    },
    {
      "epoch": 2.1038753495805036,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 5266
    },
    {
      "epoch": 2.104274870155813,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 5267
    },
    {
      "epoch": 2.1046743907311227,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 5268
    },
    {
      "epoch": 2.105073911306432,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 5269
    },
    {
      "epoch": 2.105473431881742,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 5270
    },
    {
      "epoch": 2.1058729524570516,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 5271
    },
    {
      "epoch": 2.106272473032361,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 5272
    },
    {
      "epoch": 2.1066719936076708,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 5273
    },
    {
      "epoch": 2.1070715141829806,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 5274
    },
    {
      "epoch": 2.10747103475829,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 5275
    },
    {
      "epoch": 2.1078705553335997,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 5276
    },
    {
      "epoch": 2.1082700759089095,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 5277
    },
    {
      "epoch": 2.108669596484219,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 5278
    },
    {
      "epoch": 2.1090691170595286,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 5279
    },
    {
      "epoch": 2.1094686376348384,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 5280
    },
    {
      "epoch": 2.1098681582101477,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 5281
    },
    {
      "epoch": 2.1102676787854575,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0741,
      "step": 5282
    },
    {
      "epoch": 2.1106671993607673,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 5283
    },
    {
      "epoch": 2.1110667199360766,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0716,
      "step": 5284
    },
    {
      "epoch": 2.1114662405113864,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 5285
    },
    {
      "epoch": 2.1118657610866958,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 5286
    },
    {
      "epoch": 2.1122652816620056,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 5287
    },
    {
      "epoch": 2.1126648022373153,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 5288
    },
    {
      "epoch": 2.1130643228126247,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 5289
    },
    {
      "epoch": 2.1134638433879345,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 5290
    },
    {
      "epoch": 2.1138633639632443,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 5291
    },
    {
      "epoch": 2.1142628845385536,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 5292
    },
    {
      "epoch": 2.1146624051138634,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 5293
    },
    {
      "epoch": 2.115061925689173,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 5294
    },
    {
      "epoch": 2.1154614462644825,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 5295
    },
    {
      "epoch": 2.1158609668397923,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 5296
    },
    {
      "epoch": 2.116260487415102,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 5297
    },
    {
      "epoch": 2.1166600079904114,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 5298
    },
    {
      "epoch": 2.117059528565721,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 5299
    },
    {
      "epoch": 2.1174590491410306,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0717,
      "step": 5300
    },
    {
      "epoch": 2.1178585697163403,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 5301
    },
    {
      "epoch": 2.11825809029165,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 5302
    },
    {
      "epoch": 2.1186576108669595,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 5303
    },
    {
      "epoch": 2.1190571314422693,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 5304
    },
    {
      "epoch": 2.119456652017579,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.076,
      "step": 5305
    },
    {
      "epoch": 2.1198561725928884,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 5306
    },
    {
      "epoch": 2.120255693168198,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 5307
    },
    {
      "epoch": 2.120655213743508,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 5308
    },
    {
      "epoch": 2.1210547343188173,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 5309
    },
    {
      "epoch": 2.121454254894127,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 5310
    },
    {
      "epoch": 2.121853775469437,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 5311
    },
    {
      "epoch": 2.122253296044746,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 5312
    },
    {
      "epoch": 2.122652816620056,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0715,
      "step": 5313
    },
    {
      "epoch": 2.1230523371953653,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 5314
    },
    {
      "epoch": 2.123451857770675,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 5315
    },
    {
      "epoch": 2.123851378345985,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 5316
    },
    {
      "epoch": 2.1242508989212943,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 5317
    },
    {
      "epoch": 2.124650419496604,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 5318
    },
    {
      "epoch": 2.125049940071914,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 5319
    },
    {
      "epoch": 2.125449460647223,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 5320
    },
    {
      "epoch": 2.125848981222533,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 5321
    },
    {
      "epoch": 2.1262485017978427,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0753,
      "step": 5322
    },
    {
      "epoch": 2.126648022373152,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 5323
    },
    {
      "epoch": 2.127047542948462,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 5324
    },
    {
      "epoch": 2.1274470635237717,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 5325
    },
    {
      "epoch": 2.127846584099081,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 5326
    },
    {
      "epoch": 2.128246104674391,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 5327
    },
    {
      "epoch": 2.1286456252497006,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 5328
    },
    {
      "epoch": 2.12904514582501,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 5329
    },
    {
      "epoch": 2.1294446664003197,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 5330
    },
    {
      "epoch": 2.129844186975629,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 5331
    },
    {
      "epoch": 2.130243707550939,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 5332
    },
    {
      "epoch": 2.1306432281262486,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 5333
    },
    {
      "epoch": 2.131042748701558,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 5334
    },
    {
      "epoch": 2.1314422692768678,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 5335
    },
    {
      "epoch": 2.1318417898521775,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 5336
    },
    {
      "epoch": 2.132241310427487,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 5337
    },
    {
      "epoch": 2.1326408310027967,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 5338
    },
    {
      "epoch": 2.1330403515781065,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 5339
    },
    {
      "epoch": 2.133439872153416,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 5340
    },
    {
      "epoch": 2.1338393927287256,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 5341
    },
    {
      "epoch": 2.134238913304035,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 5342
    },
    {
      "epoch": 2.1346384338793447,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0783,
      "step": 5343
    },
    {
      "epoch": 2.1350379544546545,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 5344
    },
    {
      "epoch": 2.135437475029964,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 5345
    },
    {
      "epoch": 2.1358369956052736,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 5346
    },
    {
      "epoch": 2.1362365161805834,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 5347
    },
    {
      "epoch": 2.1366360367558928,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 5348
    },
    {
      "epoch": 2.1370355573312025,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 5349
    },
    {
      "epoch": 2.1374350779065123,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 5350
    },
    {
      "epoch": 2.1378345984818217,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 5351
    },
    {
      "epoch": 2.1382341190571315,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 5352
    },
    {
      "epoch": 2.1386336396324412,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 5353
    },
    {
      "epoch": 2.1390331602077506,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0745,
      "step": 5354
    },
    {
      "epoch": 2.1394326807830604,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 5355
    },
    {
      "epoch": 2.13983220135837,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 5356
    },
    {
      "epoch": 2.1402317219336795,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0737,
      "step": 5357
    },
    {
      "epoch": 2.1406312425089893,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 5358
    },
    {
      "epoch": 2.1410307630842986,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 5359
    },
    {
      "epoch": 2.1414302836596084,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 5360
    },
    {
      "epoch": 2.141829804234918,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 5361
    },
    {
      "epoch": 2.1422293248102275,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 5362
    },
    {
      "epoch": 2.1426288453855373,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 5363
    },
    {
      "epoch": 2.143028365960847,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 5364
    },
    {
      "epoch": 2.1434278865361565,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 5365
    },
    {
      "epoch": 2.1438274071114662,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 5366
    },
    {
      "epoch": 2.144226927686776,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 5367
    },
    {
      "epoch": 2.1446264482620854,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 5368
    },
    {
      "epoch": 2.145025968837395,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 5369
    },
    {
      "epoch": 2.145425489412705,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 5370
    },
    {
      "epoch": 2.1458250099880143,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 5371
    },
    {
      "epoch": 2.146224530563324,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 5372
    },
    {
      "epoch": 2.146624051138634,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 5373
    },
    {
      "epoch": 2.147023571713943,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 5374
    },
    {
      "epoch": 2.147423092289253,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 5375
    },
    {
      "epoch": 2.1478226128645623,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.074,
      "step": 5376
    },
    {
      "epoch": 2.148222133439872,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 5377
    },
    {
      "epoch": 2.148621654015182,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 5378
    },
    {
      "epoch": 2.1490211745904912,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 5379
    },
    {
      "epoch": 2.149420695165801,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 5380
    },
    {
      "epoch": 2.149820215741111,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 5381
    },
    {
      "epoch": 2.15021973631642,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 5382
    },
    {
      "epoch": 2.15061925689173,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 5383
    },
    {
      "epoch": 2.1510187774670397,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 5384
    },
    {
      "epoch": 2.151418298042349,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 5385
    },
    {
      "epoch": 2.151817818617659,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 5386
    },
    {
      "epoch": 2.152217339192968,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 5387
    },
    {
      "epoch": 2.152616859768278,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 5388
    },
    {
      "epoch": 2.1530163803435878,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 5389
    },
    {
      "epoch": 2.153415900918897,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0693,
      "step": 5390
    },
    {
      "epoch": 2.153815421494207,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 5391
    },
    {
      "epoch": 2.1542149420695167,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 5392
    },
    {
      "epoch": 2.154614462644826,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 5393
    },
    {
      "epoch": 2.155013983220136,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 5394
    },
    {
      "epoch": 2.1554135037954456,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 5395
    },
    {
      "epoch": 2.155813024370755,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 5396
    },
    {
      "epoch": 2.1562125449460647,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 5397
    },
    {
      "epoch": 2.1566120655213745,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 5398
    },
    {
      "epoch": 2.157011586096684,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 5399
    },
    {
      "epoch": 2.1574111066719937,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 5400
    },
    {
      "epoch": 2.1578106272473034,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 5401
    },
    {
      "epoch": 2.158210147822613,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 5402
    },
    {
      "epoch": 2.1586096683979226,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 5403
    },
    {
      "epoch": 2.159009188973232,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 5404
    },
    {
      "epoch": 2.1594087095485417,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 5405
    },
    {
      "epoch": 2.1598082301238515,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 5406
    },
    {
      "epoch": 2.160207750699161,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0712,
      "step": 5407
    },
    {
      "epoch": 2.1606072712744706,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 5408
    },
    {
      "epoch": 2.1610067918497804,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 5409
    },
    {
      "epoch": 2.1614063124250897,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 5410
    },
    {
      "epoch": 2.1618058330003995,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 5411
    },
    {
      "epoch": 2.1622053535757093,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 5412
    },
    {
      "epoch": 2.1626048741510187,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 5413
    },
    {
      "epoch": 2.1630043947263284,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 5414
    },
    {
      "epoch": 2.1634039153016382,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 5415
    },
    {
      "epoch": 2.1638034358769476,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 5416
    },
    {
      "epoch": 2.1642029564522574,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 5417
    },
    {
      "epoch": 2.164602477027567,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 5418
    },
    {
      "epoch": 2.1650019976028765,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 5419
    },
    {
      "epoch": 2.1654015181781863,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 5420
    },
    {
      "epoch": 2.1658010387534956,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 5421
    },
    {
      "epoch": 2.1662005593288054,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.07,
      "step": 5422
    },
    {
      "epoch": 2.166600079904115,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 5423
    },
    {
      "epoch": 2.1669996004794245,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 5424
    },
    {
      "epoch": 2.1673991210547343,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 5425
    },
    {
      "epoch": 2.167798641630044,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 5426
    },
    {
      "epoch": 2.1681981622053534,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 5427
    },
    {
      "epoch": 2.1685976827806632,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 5428
    },
    {
      "epoch": 2.168997203355973,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 5429
    },
    {
      "epoch": 2.1693967239312824,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 5430
    },
    {
      "epoch": 2.169796244506592,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0743,
      "step": 5431
    },
    {
      "epoch": 2.1701957650819015,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 5432
    },
    {
      "epoch": 2.1705952856572113,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0719,
      "step": 5433
    },
    {
      "epoch": 2.170994806232521,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0687,
      "step": 5434
    },
    {
      "epoch": 2.1713943268078304,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 5435
    },
    {
      "epoch": 2.17179384738314,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 5436
    },
    {
      "epoch": 2.17219336795845,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 5437
    },
    {
      "epoch": 2.1725928885337593,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 5438
    },
    {
      "epoch": 2.172992409109069,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 5439
    },
    {
      "epoch": 2.173391929684379,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 5440
    },
    {
      "epoch": 2.1737914502596882,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 5441
    },
    {
      "epoch": 2.174190970834998,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 5442
    },
    {
      "epoch": 2.174590491410308,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 5443
    },
    {
      "epoch": 2.174990011985617,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 5444
    },
    {
      "epoch": 2.175389532560927,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 5445
    },
    {
      "epoch": 2.1757890531362367,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 5446
    },
    {
      "epoch": 2.176188573711546,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 5447
    },
    {
      "epoch": 2.176588094286856,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 5448
    },
    {
      "epoch": 2.176987614862165,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 5449
    },
    {
      "epoch": 2.177387135437475,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 5450
    },
    {
      "epoch": 2.1777866560127848,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 5451
    },
    {
      "epoch": 2.178186176588094,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 5452
    },
    {
      "epoch": 2.178585697163404,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 5453
    },
    {
      "epoch": 2.1789852177387137,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 5454
    },
    {
      "epoch": 2.179384738314023,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0741,
      "step": 5455
    },
    {
      "epoch": 2.179784258889333,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 5456
    },
    {
      "epoch": 2.1801837794646426,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 5457
    },
    {
      "epoch": 2.180583300039952,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 5458
    },
    {
      "epoch": 2.1809828206152617,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 5459
    },
    {
      "epoch": 2.1813823411905715,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 5460
    },
    {
      "epoch": 2.181781861765881,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 5461
    },
    {
      "epoch": 2.1821813823411906,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 5462
    },
    {
      "epoch": 2.1825809029165004,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 5463
    },
    {
      "epoch": 2.1829804234918098,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 5464
    },
    {
      "epoch": 2.1833799440671195,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 5465
    },
    {
      "epoch": 2.183779464642429,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 5466
    },
    {
      "epoch": 2.1841789852177387,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 5467
    },
    {
      "epoch": 2.1845785057930485,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 5468
    },
    {
      "epoch": 2.184978026368358,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 5469
    },
    {
      "epoch": 2.1853775469436676,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0715,
      "step": 5470
    },
    {
      "epoch": 2.1857770675189774,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 5471
    },
    {
      "epoch": 2.1861765880942867,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 5472
    },
    {
      "epoch": 2.1865761086695965,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 5473
    },
    {
      "epoch": 2.1869756292449063,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 5474
    },
    {
      "epoch": 2.1873751498202156,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 5475
    },
    {
      "epoch": 2.1877746703955254,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0755,
      "step": 5476
    },
    {
      "epoch": 2.1881741909708348,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 5477
    },
    {
      "epoch": 2.1885737115461446,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 5478
    },
    {
      "epoch": 2.1889732321214543,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 5479
    },
    {
      "epoch": 2.1893727526967637,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 5480
    },
    {
      "epoch": 2.1897722732720735,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0703,
      "step": 5481
    },
    {
      "epoch": 2.1901717938473833,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 5482
    },
    {
      "epoch": 2.1905713144226926,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 5483
    },
    {
      "epoch": 2.1909708349980024,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 5484
    },
    {
      "epoch": 2.191370355573312,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 5485
    },
    {
      "epoch": 2.1917698761486215,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 5486
    },
    {
      "epoch": 2.1921693967239313,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 5487
    },
    {
      "epoch": 2.192568917299241,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 5488
    },
    {
      "epoch": 2.1929684378745504,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 5489
    },
    {
      "epoch": 2.19336795844986,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 5490
    },
    {
      "epoch": 2.19376747902517,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 5491
    },
    {
      "epoch": 2.1941669996004793,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 5492
    },
    {
      "epoch": 2.194566520175789,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 5493
    },
    {
      "epoch": 2.1949660407510985,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 5494
    },
    {
      "epoch": 2.1953655613264083,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 5495
    },
    {
      "epoch": 2.195765081901718,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 5496
    },
    {
      "epoch": 2.1961646024770274,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 5497
    },
    {
      "epoch": 2.196564123052337,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 5498
    },
    {
      "epoch": 2.196963643627647,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 5499
    },
    {
      "epoch": 2.1973631642029563,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 5500
    },
    {
      "epoch": 2.197762684778266,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 5501
    },
    {
      "epoch": 2.198162205353576,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 5502
    },
    {
      "epoch": 2.198561725928885,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 5503
    },
    {
      "epoch": 2.198961246504195,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 5504
    },
    {
      "epoch": 2.199360767079505,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 5505
    },
    {
      "epoch": 2.199760287654814,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 5506
    },
    {
      "epoch": 2.200159808230124,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 5507
    },
    {
      "epoch": 2.2005593288054337,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 5508
    },
    {
      "epoch": 2.200958849380743,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 5509
    },
    {
      "epoch": 2.201358369956053,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 5510
    },
    {
      "epoch": 2.201757890531362,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 5511
    },
    {
      "epoch": 2.202157411106672,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 5512
    },
    {
      "epoch": 2.2025569316819817,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 5513
    },
    {
      "epoch": 2.202956452257291,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 5514
    },
    {
      "epoch": 2.203355972832601,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0651,
      "step": 5515
    },
    {
      "epoch": 2.2037554934079107,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 5516
    },
    {
      "epoch": 2.20415501398322,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 5517
    },
    {
      "epoch": 2.20455453455853,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 5518
    },
    {
      "epoch": 2.2049540551338396,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 5519
    },
    {
      "epoch": 2.205353575709149,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 5520
    },
    {
      "epoch": 2.2057530962844587,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 5521
    },
    {
      "epoch": 2.206152616859768,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 5522
    },
    {
      "epoch": 2.206552137435078,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 5523
    },
    {
      "epoch": 2.2069516580103876,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 5524
    },
    {
      "epoch": 2.207351178585697,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 5525
    },
    {
      "epoch": 2.2077506991610067,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 5526
    },
    {
      "epoch": 2.2081502197363165,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 5527
    },
    {
      "epoch": 2.208549740311626,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 5528
    },
    {
      "epoch": 2.2089492608869357,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 5529
    },
    {
      "epoch": 2.2093487814622454,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 5530
    },
    {
      "epoch": 2.209748302037555,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 5531
    },
    {
      "epoch": 2.2101478226128646,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 5532
    },
    {
      "epoch": 2.2105473431881744,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 5533
    },
    {
      "epoch": 2.2109468637634837,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 5534
    },
    {
      "epoch": 2.2113463843387935,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 5535
    },
    {
      "epoch": 2.2117459049141033,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 5536
    },
    {
      "epoch": 2.2121454254894126,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0687,
      "step": 5537
    },
    {
      "epoch": 2.2125449460647224,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 5538
    },
    {
      "epoch": 2.2129444666400317,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 5539
    },
    {
      "epoch": 2.2133439872153415,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 5540
    },
    {
      "epoch": 2.2137435077906513,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 5541
    },
    {
      "epoch": 2.2141430283659607,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 5542
    },
    {
      "epoch": 2.2145425489412705,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 5543
    },
    {
      "epoch": 2.2149420695165802,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 5544
    },
    {
      "epoch": 2.2153415900918896,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 5545
    },
    {
      "epoch": 2.2157411106671994,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 5546
    },
    {
      "epoch": 2.216140631242509,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 5547
    },
    {
      "epoch": 2.2165401518178185,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 5548
    },
    {
      "epoch": 2.2169396723931283,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 5549
    },
    {
      "epoch": 2.217339192968438,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 5550
    },
    {
      "epoch": 2.2177387135437474,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 5551
    },
    {
      "epoch": 2.218138234119057,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 5552
    },
    {
      "epoch": 2.218537754694367,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 5553
    },
    {
      "epoch": 2.2189372752696763,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0758,
      "step": 5554
    },
    {
      "epoch": 2.219336795844986,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 5555
    },
    {
      "epoch": 2.2197363164202955,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 5556
    },
    {
      "epoch": 2.2201358369956052,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 5557
    },
    {
      "epoch": 2.220535357570915,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 5558
    },
    {
      "epoch": 2.2209348781462244,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 5559
    },
    {
      "epoch": 2.221334398721534,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 5560
    },
    {
      "epoch": 2.221733919296844,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 5561
    },
    {
      "epoch": 2.2221334398721533,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 5562
    },
    {
      "epoch": 2.222532960447463,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 5563
    },
    {
      "epoch": 2.222932481022773,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 5564
    },
    {
      "epoch": 2.223332001598082,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 5565
    },
    {
      "epoch": 2.223731522173392,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 5566
    },
    {
      "epoch": 2.2241310427487013,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 5567
    },
    {
      "epoch": 2.224530563324011,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 5568
    },
    {
      "epoch": 2.224930083899321,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 5569
    },
    {
      "epoch": 2.2253296044746302,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 5570
    },
    {
      "epoch": 2.22572912504994,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 5571
    },
    {
      "epoch": 2.22612864562525,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 5572
    },
    {
      "epoch": 2.226528166200559,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 5573
    },
    {
      "epoch": 2.226927686775869,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 5574
    },
    {
      "epoch": 2.2273272073511787,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0665,
      "step": 5575
    },
    {
      "epoch": 2.227726727926488,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 5576
    },
    {
      "epoch": 2.228126248501798,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 5577
    },
    {
      "epoch": 2.2285257690771076,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 5578
    },
    {
      "epoch": 2.228925289652417,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 5579
    },
    {
      "epoch": 2.2293248102277268,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 5580
    },
    {
      "epoch": 2.2297243308030366,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 5581
    },
    {
      "epoch": 2.230123851378346,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 5582
    },
    {
      "epoch": 2.2305233719536557,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 5583
    },
    {
      "epoch": 2.230922892528965,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 5584
    },
    {
      "epoch": 2.231322413104275,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0737,
      "step": 5585
    },
    {
      "epoch": 2.2317219336795846,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 5586
    },
    {
      "epoch": 2.232121454254894,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 5587
    },
    {
      "epoch": 2.2325209748302037,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 5588
    },
    {
      "epoch": 2.2329204954055135,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 5589
    },
    {
      "epoch": 2.233320015980823,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 5590
    },
    {
      "epoch": 2.2337195365561326,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 5591
    },
    {
      "epoch": 2.2341190571314424,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 5592
    },
    {
      "epoch": 2.2345185777067518,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0805,
      "step": 5593
    },
    {
      "epoch": 2.2349180982820616,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 5594
    },
    {
      "epoch": 2.2353176188573713,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 5595
    },
    {
      "epoch": 2.2357171394326807,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 5596
    },
    {
      "epoch": 2.2361166600079905,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 5597
    },
    {
      "epoch": 2.2365161805833003,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 5598
    },
    {
      "epoch": 2.2369157011586096,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 5599
    },
    {
      "epoch": 2.2373152217339194,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0651,
      "step": 5600
    },
    {
      "epoch": 2.2377147423092287,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 5601
    },
    {
      "epoch": 2.2381142628845385,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0693,
      "step": 5602
    },
    {
      "epoch": 2.2385137834598483,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 5603
    },
    {
      "epoch": 2.2389133040351576,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 5604
    },
    {
      "epoch": 2.2393128246104674,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 5605
    },
    {
      "epoch": 2.239712345185777,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 5606
    },
    {
      "epoch": 2.2401118657610866,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 5607
    },
    {
      "epoch": 2.2405113863363963,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 5608
    },
    {
      "epoch": 2.240910906911706,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 5609
    },
    {
      "epoch": 2.2413104274870155,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 5610
    },
    {
      "epoch": 2.2417099480623253,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 5611
    },
    {
      "epoch": 2.2421094686376346,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 5612
    },
    {
      "epoch": 2.2425089892129444,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 5613
    },
    {
      "epoch": 2.242908509788254,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 5614
    },
    {
      "epoch": 2.2433080303635635,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 5615
    },
    {
      "epoch": 2.2437075509388733,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 5616
    },
    {
      "epoch": 2.244107071514183,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 5617
    },
    {
      "epoch": 2.2445065920894924,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0693,
      "step": 5618
    },
    {
      "epoch": 2.2449061126648022,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 5619
    },
    {
      "epoch": 2.245305633240112,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 5620
    },
    {
      "epoch": 2.2457051538154214,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 5621
    },
    {
      "epoch": 2.246104674390731,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 5622
    },
    {
      "epoch": 2.246504194966041,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 5623
    },
    {
      "epoch": 2.2469037155413503,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 5624
    },
    {
      "epoch": 2.24730323611666,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 5625
    },
    {
      "epoch": 2.24770275669197,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 5626
    },
    {
      "epoch": 2.248102277267279,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 5627
    },
    {
      "epoch": 2.248501797842589,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 5628
    },
    {
      "epoch": 2.2489013184178983,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 5629
    },
    {
      "epoch": 2.249300838993208,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 5630
    },
    {
      "epoch": 2.249700359568518,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 5631
    },
    {
      "epoch": 2.2500998801438272,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 5632
    },
    {
      "epoch": 2.250499400719137,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 5633
    },
    {
      "epoch": 2.250898921294447,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 5634
    },
    {
      "epoch": 2.251298441869756,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 5635
    },
    {
      "epoch": 2.251697962445066,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 5636
    },
    {
      "epoch": 2.2520974830203757,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 5637
    },
    {
      "epoch": 2.252497003595685,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 5638
    },
    {
      "epoch": 2.252896524170995,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 5639
    },
    {
      "epoch": 2.2532960447463046,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 5640
    },
    {
      "epoch": 2.253695565321614,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 5641
    },
    {
      "epoch": 2.2540950858969238,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 5642
    },
    {
      "epoch": 2.2544946064722335,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 5643
    },
    {
      "epoch": 2.254894127047543,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 5644
    },
    {
      "epoch": 2.2552936476228527,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 5645
    },
    {
      "epoch": 2.255693168198162,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 5646
    },
    {
      "epoch": 2.256092688773472,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 5647
    },
    {
      "epoch": 2.2564922093487816,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 5648
    },
    {
      "epoch": 2.256891729924091,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 5649
    },
    {
      "epoch": 2.2572912504994007,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0704,
      "step": 5650
    },
    {
      "epoch": 2.2576907710747105,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0753,
      "step": 5651
    },
    {
      "epoch": 2.25809029165002,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 5652
    },
    {
      "epoch": 2.2584898122253296,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0768,
      "step": 5653
    },
    {
      "epoch": 2.2588893328006394,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 5654
    },
    {
      "epoch": 2.2592888533759488,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 5655
    },
    {
      "epoch": 2.2596883739512585,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 5656
    },
    {
      "epoch": 2.260087894526568,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 5657
    },
    {
      "epoch": 2.2604874151018777,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 5658
    },
    {
      "epoch": 2.2608869356771875,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 5659
    },
    {
      "epoch": 2.2612864562524972,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 5660
    },
    {
      "epoch": 2.2616859768278066,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 5661
    },
    {
      "epoch": 2.2620854974031164,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 5662
    },
    {
      "epoch": 2.2624850179784257,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0781,
      "step": 5663
    },
    {
      "epoch": 2.2628845385537355,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 5664
    },
    {
      "epoch": 2.2632840591290453,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 5665
    },
    {
      "epoch": 2.2636835797043546,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 5666
    },
    {
      "epoch": 2.2640831002796644,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 5667
    },
    {
      "epoch": 2.264482620854974,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 5668
    },
    {
      "epoch": 2.2648821414302835,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 5669
    },
    {
      "epoch": 2.2652816620055933,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 5670
    },
    {
      "epoch": 2.265681182580903,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 5671
    },
    {
      "epoch": 2.2660807031562125,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 5672
    },
    {
      "epoch": 2.2664802237315222,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 5673
    },
    {
      "epoch": 2.2668797443068316,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 5674
    },
    {
      "epoch": 2.2672792648821414,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 5675
    },
    {
      "epoch": 2.267678785457451,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0728,
      "step": 5676
    },
    {
      "epoch": 2.2680783060327605,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 5677
    },
    {
      "epoch": 2.2684778266080703,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 5678
    },
    {
      "epoch": 2.26887734718338,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0687,
      "step": 5679
    },
    {
      "epoch": 2.2692768677586894,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 5680
    },
    {
      "epoch": 2.269676388333999,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 5681
    },
    {
      "epoch": 2.270075908909309,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 5682
    },
    {
      "epoch": 2.2704754294846183,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 5683
    },
    {
      "epoch": 2.270874950059928,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 5684
    },
    {
      "epoch": 2.271274470635238,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0665,
      "step": 5685
    },
    {
      "epoch": 2.2716739912105472,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 5686
    },
    {
      "epoch": 2.272073511785857,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 5687
    },
    {
      "epoch": 2.272473032361167,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 5688
    },
    {
      "epoch": 2.272872552936476,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 5689
    },
    {
      "epoch": 2.273272073511786,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 5690
    },
    {
      "epoch": 2.2736715940870953,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 5691
    },
    {
      "epoch": 2.274071114662405,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 5692
    },
    {
      "epoch": 2.274470635237715,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 5693
    },
    {
      "epoch": 2.274870155813024,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 5694
    },
    {
      "epoch": 2.275269676388334,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 5695
    },
    {
      "epoch": 2.275669196963644,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0739,
      "step": 5696
    },
    {
      "epoch": 2.276068717538953,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0657,
      "step": 5697
    },
    {
      "epoch": 2.276468238114263,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 5698
    },
    {
      "epoch": 2.2768677586895727,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 5699
    },
    {
      "epoch": 2.277267279264882,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0736,
      "step": 5700
    },
    {
      "epoch": 2.277666799840192,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 5701
    },
    {
      "epoch": 2.278066320415501,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 5702
    },
    {
      "epoch": 2.278465840990811,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 5703
    },
    {
      "epoch": 2.2788653615661207,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 5704
    },
    {
      "epoch": 2.2792648821414305,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 5705
    },
    {
      "epoch": 2.27966440271674,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 5706
    },
    {
      "epoch": 2.2800639232920497,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 5707
    },
    {
      "epoch": 2.280463443867359,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 5708
    },
    {
      "epoch": 2.280862964442669,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 5709
    },
    {
      "epoch": 2.2812624850179786,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 5710
    },
    {
      "epoch": 2.281662005593288,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 5711
    },
    {
      "epoch": 2.2820615261685977,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 5712
    },
    {
      "epoch": 2.2824610467439075,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 5713
    },
    {
      "epoch": 2.282860567319217,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 5714
    },
    {
      "epoch": 2.2832600878945266,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 5715
    },
    {
      "epoch": 2.2836596084698364,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 5716
    },
    {
      "epoch": 2.2840591290451457,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 5717
    },
    {
      "epoch": 2.2844586496204555,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 5718
    },
    {
      "epoch": 2.284858170195765,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 5719
    },
    {
      "epoch": 2.2852576907710747,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 5720
    },
    {
      "epoch": 2.2856572113463844,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 5721
    },
    {
      "epoch": 2.286056731921694,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 5722
    },
    {
      "epoch": 2.2864562524970036,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 5723
    },
    {
      "epoch": 2.2868557730723134,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 5724
    },
    {
      "epoch": 2.2872552936476227,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 5725
    },
    {
      "epoch": 2.2876548142229325,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 5726
    },
    {
      "epoch": 2.2880543347982423,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 5727
    },
    {
      "epoch": 2.2884538553735516,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0712,
      "step": 5728
    },
    {
      "epoch": 2.2888533759488614,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 5729
    },
    {
      "epoch": 2.289252896524171,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 5730
    },
    {
      "epoch": 2.2896524170994805,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 5731
    },
    {
      "epoch": 2.2900519376747903,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 5732
    },
    {
      "epoch": 2.2904514582501,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 5733
    },
    {
      "epoch": 2.2908509788254094,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 5734
    },
    {
      "epoch": 2.2912504994007192,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 5735
    },
    {
      "epoch": 2.2916500199760286,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 5736
    },
    {
      "epoch": 2.2920495405513384,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 5737
    },
    {
      "epoch": 2.292449061126648,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 5738
    },
    {
      "epoch": 2.2928485817019575,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 5739
    },
    {
      "epoch": 2.2932481022772673,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 5740
    },
    {
      "epoch": 2.293647622852577,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 5741
    },
    {
      "epoch": 2.2940471434278864,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 5742
    },
    {
      "epoch": 2.294446664003196,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 5743
    },
    {
      "epoch": 2.294846184578506,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 5744
    },
    {
      "epoch": 2.2952457051538153,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 5745
    },
    {
      "epoch": 2.295645225729125,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 5746
    },
    {
      "epoch": 2.2960447463044344,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0598,
      "step": 5747
    },
    {
      "epoch": 2.2964442668797442,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 5748
    },
    {
      "epoch": 2.296843787455054,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 5749
    },
    {
      "epoch": 2.297243308030364,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 5750
    },
    {
      "epoch": 2.297642828605673,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0713,
      "step": 5751
    },
    {
      "epoch": 2.298042349180983,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 5752
    },
    {
      "epoch": 2.2984418697562923,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 5753
    },
    {
      "epoch": 2.298841390331602,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 5754
    },
    {
      "epoch": 2.299240910906912,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 5755
    },
    {
      "epoch": 2.299640431482221,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 5756
    },
    {
      "epoch": 2.300039952057531,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 5757
    },
    {
      "epoch": 2.3004394726328408,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 5758
    },
    {
      "epoch": 2.30083899320815,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 5759
    },
    {
      "epoch": 2.30123851378346,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 5760
    },
    {
      "epoch": 2.3016380343587697,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 5761
    },
    {
      "epoch": 2.302037554934079,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 5762
    },
    {
      "epoch": 2.302437075509389,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 5763
    },
    {
      "epoch": 2.302836596084698,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 5764
    },
    {
      "epoch": 2.303236116660008,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 5765
    },
    {
      "epoch": 2.3036356372353177,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 5766
    },
    {
      "epoch": 2.304035157810627,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 5767
    },
    {
      "epoch": 2.304434678385937,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 5768
    },
    {
      "epoch": 2.3048341989612466,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 5769
    },
    {
      "epoch": 2.305233719536556,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 5770
    },
    {
      "epoch": 2.3056332401118658,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 5771
    },
    {
      "epoch": 2.3060327606871756,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 5772
    },
    {
      "epoch": 2.306432281262485,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 5773
    },
    {
      "epoch": 2.3068318018377947,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 5774
    },
    {
      "epoch": 2.3072313224131045,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 5775
    },
    {
      "epoch": 2.307630842988414,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 5776
    },
    {
      "epoch": 2.3080303635637236,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 5777
    },
    {
      "epoch": 2.3084298841390334,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 5778
    },
    {
      "epoch": 2.3088294047143427,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 5779
    },
    {
      "epoch": 2.3092289252896525,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 5780
    },
    {
      "epoch": 2.309628445864962,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 5781
    },
    {
      "epoch": 2.3100279664402716,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 5782
    },
    {
      "epoch": 2.3104274870155814,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 5783
    },
    {
      "epoch": 2.3108270075908908,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 5784
    },
    {
      "epoch": 2.3112265281662006,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 5785
    },
    {
      "epoch": 2.3116260487415103,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 5786
    },
    {
      "epoch": 2.3120255693168197,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 5787
    },
    {
      "epoch": 2.3124250898921295,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 5788
    },
    {
      "epoch": 2.3128246104674393,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 5789
    },
    {
      "epoch": 2.3132241310427486,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 5790
    },
    {
      "epoch": 2.3136236516180584,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 5791
    },
    {
      "epoch": 2.3140231721933677,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0679,
      "step": 5792
    },
    {
      "epoch": 2.3144226927686775,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 5793
    },
    {
      "epoch": 2.3148222133439873,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 5794
    },
    {
      "epoch": 2.315221733919297,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 5795
    },
    {
      "epoch": 2.3156212544946064,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 5796
    },
    {
      "epoch": 2.316020775069916,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 5797
    },
    {
      "epoch": 2.3164202956452256,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 5798
    },
    {
      "epoch": 2.3168198162205353,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 5799
    },
    {
      "epoch": 2.317219336795845,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 5800
    },
    {
      "epoch": 2.3176188573711545,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 5801
    },
    {
      "epoch": 2.3180183779464643,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 5802
    },
    {
      "epoch": 2.318417898521774,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 5803
    },
    {
      "epoch": 2.3188174190970834,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 5804
    },
    {
      "epoch": 2.319216939672393,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 5805
    },
    {
      "epoch": 2.319616460247703,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 5806
    },
    {
      "epoch": 2.3200159808230123,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0728,
      "step": 5807
    },
    {
      "epoch": 2.320415501398322,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 5808
    },
    {
      "epoch": 2.3208150219736314,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 5809
    },
    {
      "epoch": 2.321214542548941,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 5810
    },
    {
      "epoch": 2.321614063124251,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 5811
    },
    {
      "epoch": 2.3220135836995603,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 5812
    },
    {
      "epoch": 2.32241310427487,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 5813
    },
    {
      "epoch": 2.32281262485018,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 5814
    },
    {
      "epoch": 2.3232121454254893,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 5815
    },
    {
      "epoch": 2.323611666000799,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 5816
    },
    {
      "epoch": 2.324011186576109,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0711,
      "step": 5817
    },
    {
      "epoch": 2.324410707151418,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 5818
    },
    {
      "epoch": 2.324810227726728,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 5819
    },
    {
      "epoch": 2.3252097483020377,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 5820
    },
    {
      "epoch": 2.325609268877347,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 5821
    },
    {
      "epoch": 2.326008789452657,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 5822
    },
    {
      "epoch": 2.3264083100279667,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 5823
    },
    {
      "epoch": 2.326807830603276,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 5824
    },
    {
      "epoch": 2.327207351178586,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 5825
    },
    {
      "epoch": 2.327606871753895,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 5826
    },
    {
      "epoch": 2.328006392329205,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 5827
    },
    {
      "epoch": 2.3284059129045147,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 5828
    },
    {
      "epoch": 2.328805433479824,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 5829
    },
    {
      "epoch": 2.329204954055134,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 5830
    },
    {
      "epoch": 2.3296044746304436,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 5831
    },
    {
      "epoch": 2.330003995205753,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 5832
    },
    {
      "epoch": 2.3304035157810628,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 5833
    },
    {
      "epoch": 2.3308030363563725,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 5834
    },
    {
      "epoch": 2.331202556931682,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 5835
    },
    {
      "epoch": 2.3316020775069917,
      "grad_norm": 0.25,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 5836
    },
    {
      "epoch": 2.332001598082301,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 5837
    },
    {
      "epoch": 2.332401118657611,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 5838
    },
    {
      "epoch": 2.3328006392329206,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 5839
    },
    {
      "epoch": 2.3332001598082304,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 5840
    },
    {
      "epoch": 2.3335996803835397,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 5841
    },
    {
      "epoch": 2.3339992009588495,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 5842
    },
    {
      "epoch": 2.334398721534159,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 5843
    },
    {
      "epoch": 2.3347982421094686,
      "grad_norm": 0.265625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 5844
    },
    {
      "epoch": 2.3351977626847784,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 5845
    },
    {
      "epoch": 2.3355972832600878,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0005,
      "loss": 1.0772,
      "step": 5846
    },
    {
      "epoch": 2.3359968038353975,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 5847
    },
    {
      "epoch": 2.3363963244107073,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 5848
    },
    {
      "epoch": 2.3367958449860167,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 5849
    },
    {
      "epoch": 2.3371953655613265,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 5850
    },
    {
      "epoch": 2.3375948861366362,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 5851
    },
    {
      "epoch": 2.3379944067119456,
      "grad_norm": 0.40625,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 5852
    },
    {
      "epoch": 2.3383939272872554,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 5853
    },
    {
      "epoch": 2.3387934478625647,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 5854
    },
    {
      "epoch": 2.3391929684378745,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 5855
    },
    {
      "epoch": 2.3395924890131843,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 5856
    },
    {
      "epoch": 2.3399920095884936,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 5857
    },
    {
      "epoch": 2.3403915301638034,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 5858
    },
    {
      "epoch": 2.340791050739113,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 5859
    },
    {
      "epoch": 2.3411905713144225,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 5860
    },
    {
      "epoch": 2.3415900918897323,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 5861
    },
    {
      "epoch": 2.341989612465042,
      "grad_norm": 0.375,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 5862
    },
    {
      "epoch": 2.3423891330403515,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 5863
    },
    {
      "epoch": 2.3427886536156612,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 5864
    },
    {
      "epoch": 2.343188174190971,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 5865
    },
    {
      "epoch": 2.3435876947662804,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 5866
    },
    {
      "epoch": 2.34398721534159,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 5867
    },
    {
      "epoch": 2.3443867359169,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 5868
    },
    {
      "epoch": 2.3447862564922093,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 5869
    },
    {
      "epoch": 2.345185777067519,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 5870
    },
    {
      "epoch": 2.3455852976428284,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 5871
    },
    {
      "epoch": 2.345984818218138,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005,
      "loss": 1.0754,
      "step": 5872
    },
    {
      "epoch": 2.346384338793448,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 5873
    },
    {
      "epoch": 2.3467838593687573,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0005,
      "loss": 1.0685,
      "step": 5874
    },
    {
      "epoch": 2.347183379944067,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 5875
    },
    {
      "epoch": 2.347582900519377,
      "grad_norm": 0.40625,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 5876
    },
    {
      "epoch": 2.3479824210946862,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 5877
    },
    {
      "epoch": 2.348381941669996,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 5878
    },
    {
      "epoch": 2.348781462245306,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 5879
    },
    {
      "epoch": 2.349180982820615,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 5880
    },
    {
      "epoch": 2.349580503395925,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 5881
    },
    {
      "epoch": 2.3499800239712343,
      "grad_norm": 0.609375,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 5882
    },
    {
      "epoch": 2.350379544546544,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 5883
    },
    {
      "epoch": 2.350779065121854,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 5884
    },
    {
      "epoch": 2.3511785856971636,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0005,
      "loss": 1.0696,
      "step": 5885
    },
    {
      "epoch": 2.351578106272473,
      "grad_norm": 0.53125,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 5886
    },
    {
      "epoch": 2.3519776268477828,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 5887
    },
    {
      "epoch": 2.352377147423092,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 5888
    },
    {
      "epoch": 2.352776667998402,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 5889
    },
    {
      "epoch": 2.3531761885737117,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 5890
    },
    {
      "epoch": 2.353575709149021,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 5891
    },
    {
      "epoch": 2.353975229724331,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 5892
    },
    {
      "epoch": 2.3543747502996406,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0005,
      "loss": 1.0726,
      "step": 5893
    },
    {
      "epoch": 2.35477427087495,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 5894
    },
    {
      "epoch": 2.3551737914502597,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 5895
    },
    {
      "epoch": 2.3555733120255695,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 5896
    },
    {
      "epoch": 2.355972832600879,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 5897
    },
    {
      "epoch": 2.3563723531761886,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 5898
    },
    {
      "epoch": 2.356771873751498,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 5899
    },
    {
      "epoch": 2.357171394326808,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 5900
    },
    {
      "epoch": 2.3575709149021176,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 5901
    },
    {
      "epoch": 2.357970435477427,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 5902
    },
    {
      "epoch": 2.3583699560527367,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 5903
    },
    {
      "epoch": 2.3587694766280465,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 5904
    },
    {
      "epoch": 2.359168997203356,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005,
      "loss": 1.0657,
      "step": 5905
    },
    {
      "epoch": 2.3595685177786656,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 5906
    },
    {
      "epoch": 2.3599680383539754,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 5907
    },
    {
      "epoch": 2.3603675589292847,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 5908
    },
    {
      "epoch": 2.3607670795045945,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 5909
    },
    {
      "epoch": 2.361166600079904,
      "grad_norm": 0.25,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 5910
    },
    {
      "epoch": 2.3615661206552137,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 5911
    },
    {
      "epoch": 2.3619656412305234,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 5912
    },
    {
      "epoch": 2.3623651618058332,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 5913
    },
    {
      "epoch": 2.3627646823811426,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 5914
    },
    {
      "epoch": 2.3631642029564524,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 5915
    },
    {
      "epoch": 2.3635637235317617,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 5916
    },
    {
      "epoch": 2.3639632441070715,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 5917
    },
    {
      "epoch": 2.3643627646823813,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 5918
    },
    {
      "epoch": 2.3647622852576906,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0706,
      "step": 5919
    },
    {
      "epoch": 2.3651618058330004,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 5920
    },
    {
      "epoch": 2.36556132640831,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 5921
    },
    {
      "epoch": 2.3659608469836195,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 5922
    },
    {
      "epoch": 2.3663603675589293,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 5923
    },
    {
      "epoch": 2.366759888134239,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 5924
    },
    {
      "epoch": 2.3671594087095484,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 5925
    },
    {
      "epoch": 2.3675589292848582,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 5926
    },
    {
      "epoch": 2.3679584498601676,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 5927
    },
    {
      "epoch": 2.3683579704354774,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 5928
    },
    {
      "epoch": 2.368757491010787,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 5929
    },
    {
      "epoch": 2.369157011586097,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 5930
    },
    {
      "epoch": 2.3695565321614063,
      "grad_norm": 0.25,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 5931
    },
    {
      "epoch": 2.369956052736716,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 5932
    },
    {
      "epoch": 2.3703555733120254,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 5933
    },
    {
      "epoch": 2.370755093887335,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 5934
    },
    {
      "epoch": 2.371154614462645,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 5935
    },
    {
      "epoch": 2.3715541350379543,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 5936
    },
    {
      "epoch": 2.371953655613264,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 5937
    },
    {
      "epoch": 2.372353176188574,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 5938
    },
    {
      "epoch": 2.3727526967638832,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 5939
    },
    {
      "epoch": 2.373152217339193,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 5940
    },
    {
      "epoch": 2.373551737914503,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 5941
    },
    {
      "epoch": 2.373951258489812,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 5942
    },
    {
      "epoch": 2.374350779065122,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 5943
    },
    {
      "epoch": 2.3747502996404313,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 5944
    },
    {
      "epoch": 2.375149820215741,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0733,
      "step": 5945
    },
    {
      "epoch": 2.375549340791051,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 5946
    },
    {
      "epoch": 2.37594886136636,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 5947
    },
    {
      "epoch": 2.37634838194167,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 5948
    },
    {
      "epoch": 2.3767479025169798,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 5949
    },
    {
      "epoch": 2.377147423092289,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 5950
    },
    {
      "epoch": 2.377546943667599,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 5951
    },
    {
      "epoch": 2.3779464642429087,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 5952
    },
    {
      "epoch": 2.378345984818218,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 5953
    },
    {
      "epoch": 2.378745505393528,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 5954
    },
    {
      "epoch": 2.379145025968837,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 5955
    },
    {
      "epoch": 2.379544546544147,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 5956
    },
    {
      "epoch": 2.3799440671194567,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0706,
      "step": 5957
    },
    {
      "epoch": 2.3803435876947665,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 5958
    },
    {
      "epoch": 2.380743108270076,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 5959
    },
    {
      "epoch": 2.3811426288453856,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 5960
    },
    {
      "epoch": 2.381542149420695,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 5961
    },
    {
      "epoch": 2.3819416699960048,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 5962
    },
    {
      "epoch": 2.3823411905713145,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 5963
    },
    {
      "epoch": 2.382740711146624,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 5964
    },
    {
      "epoch": 2.3831402317219337,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 5965
    },
    {
      "epoch": 2.3835397522972435,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 5966
    },
    {
      "epoch": 2.383939272872553,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 5967
    },
    {
      "epoch": 2.3843387934478626,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 5968
    },
    {
      "epoch": 2.3847383140231724,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 5969
    },
    {
      "epoch": 2.3851378345984817,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 5970
    },
    {
      "epoch": 2.3855373551737915,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 5971
    },
    {
      "epoch": 2.385936875749101,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 5972
    },
    {
      "epoch": 2.3863363963244106,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 5973
    },
    {
      "epoch": 2.3867359168997204,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 5974
    },
    {
      "epoch": 2.38713543747503,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 5975
    },
    {
      "epoch": 2.3875349580503396,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 5976
    },
    {
      "epoch": 2.3879344786256493,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 5977
    },
    {
      "epoch": 2.3883339992009587,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0693,
      "step": 5978
    },
    {
      "epoch": 2.3887335197762685,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 5979
    },
    {
      "epoch": 2.3891330403515783,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 5980
    },
    {
      "epoch": 2.3895325609268876,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 5981
    },
    {
      "epoch": 2.3899320815021974,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 5982
    },
    {
      "epoch": 2.390331602077507,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 5983
    },
    {
      "epoch": 2.3907311226528165,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 5984
    },
    {
      "epoch": 2.3911306432281263,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 5985
    },
    {
      "epoch": 2.391530163803436,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 5986
    },
    {
      "epoch": 2.3919296843787454,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 5987
    },
    {
      "epoch": 2.392329204954055,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 5988
    },
    {
      "epoch": 2.3927287255293646,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 5989
    },
    {
      "epoch": 2.3931282461046743,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 5990
    },
    {
      "epoch": 2.393527766679984,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 5991
    },
    {
      "epoch": 2.3939272872552935,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 5992
    },
    {
      "epoch": 2.3943268078306033,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 5993
    },
    {
      "epoch": 2.394726328405913,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 5994
    },
    {
      "epoch": 2.3951258489812224,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 5995
    },
    {
      "epoch": 2.395525369556532,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 5996
    },
    {
      "epoch": 2.395924890131842,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 5997
    },
    {
      "epoch": 2.3963244107071513,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 5998
    },
    {
      "epoch": 2.396723931282461,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 5999
    },
    {
      "epoch": 2.3971234518577704,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 6000
    },
    {
      "epoch": 2.39752297243308,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 6001
    },
    {
      "epoch": 2.39792249300839,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 6002
    },
    {
      "epoch": 2.3983220135837,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 6003
    },
    {
      "epoch": 2.398721534159009,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 6004
    },
    {
      "epoch": 2.399121054734319,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 6005
    },
    {
      "epoch": 2.3995205753096283,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 6006
    },
    {
      "epoch": 2.399920095884938,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 6007
    },
    {
      "epoch": 2.400319616460248,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 6008
    },
    {
      "epoch": 2.400719137035557,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 6009
    },
    {
      "epoch": 2.401118657610867,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 6010
    },
    {
      "epoch": 2.4015181781861767,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 6011
    },
    {
      "epoch": 2.401917698761486,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 6012
    },
    {
      "epoch": 2.402317219336796,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 6013
    },
    {
      "epoch": 2.4027167399121057,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 6014
    },
    {
      "epoch": 2.403116260487415,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 6015
    },
    {
      "epoch": 2.403515781062725,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 6016
    },
    {
      "epoch": 2.403915301638034,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 6017
    },
    {
      "epoch": 2.404314822213344,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 6018
    },
    {
      "epoch": 2.4047143427886537,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 6019
    },
    {
      "epoch": 2.4051138633639635,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 6020
    },
    {
      "epoch": 2.405513383939273,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 6021
    },
    {
      "epoch": 2.4059129045145826,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 6022
    },
    {
      "epoch": 2.406312425089892,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 6023
    },
    {
      "epoch": 2.4067119456652017,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 6024
    },
    {
      "epoch": 2.4071114662405115,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 6025
    },
    {
      "epoch": 2.407510986815821,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 6026
    },
    {
      "epoch": 2.4079105073911307,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 6027
    },
    {
      "epoch": 2.4083100279664404,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 6028
    },
    {
      "epoch": 2.40870954854175,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 6029
    },
    {
      "epoch": 2.4091090691170596,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 6030
    },
    {
      "epoch": 2.4095085896923694,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 6031
    },
    {
      "epoch": 2.4099081102676787,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 6032
    },
    {
      "epoch": 2.4103076308429885,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 6033
    },
    {
      "epoch": 2.410707151418298,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 6034
    },
    {
      "epoch": 2.4111066719936076,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 6035
    },
    {
      "epoch": 2.4115061925689174,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 6036
    },
    {
      "epoch": 2.4119057131442267,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 6037
    },
    {
      "epoch": 2.4123052337195365,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 6038
    },
    {
      "epoch": 2.4127047542948463,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 6039
    },
    {
      "epoch": 2.4131042748701557,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 6040
    },
    {
      "epoch": 2.4135037954454654,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 6041
    },
    {
      "epoch": 2.4139033160207752,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 6042
    },
    {
      "epoch": 2.4143028365960846,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 6043
    },
    {
      "epoch": 2.4147023571713944,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 6044
    },
    {
      "epoch": 2.4151018777467037,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 6045
    },
    {
      "epoch": 2.4155013983220135,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 6046
    },
    {
      "epoch": 2.4159009188973233,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 6047
    },
    {
      "epoch": 2.416300439472633,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0706,
      "step": 6048
    },
    {
      "epoch": 2.4166999600479424,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 6049
    },
    {
      "epoch": 2.417099480623252,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0693,
      "step": 6050
    },
    {
      "epoch": 2.4174990011985615,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 6051
    },
    {
      "epoch": 2.4178985217738713,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 6052
    },
    {
      "epoch": 2.418298042349181,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 6053
    },
    {
      "epoch": 2.4186975629244905,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 6054
    },
    {
      "epoch": 2.4190970834998002,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 6055
    },
    {
      "epoch": 2.41949660407511,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 6056
    },
    {
      "epoch": 2.4198961246504194,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 6057
    },
    {
      "epoch": 2.420295645225729,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 6058
    },
    {
      "epoch": 2.420695165801039,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 6059
    },
    {
      "epoch": 2.4210946863763483,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 6060
    },
    {
      "epoch": 2.421494206951658,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 6061
    },
    {
      "epoch": 2.4218937275269674,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 6062
    },
    {
      "epoch": 2.422293248102277,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 6063
    },
    {
      "epoch": 2.422692768677587,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 6064
    },
    {
      "epoch": 2.4230922892528968,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 6065
    },
    {
      "epoch": 2.423491809828206,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 6066
    },
    {
      "epoch": 2.423891330403516,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 6067
    },
    {
      "epoch": 2.4242908509788252,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 6068
    },
    {
      "epoch": 2.424690371554135,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 6069
    },
    {
      "epoch": 2.425089892129445,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 6070
    },
    {
      "epoch": 2.425489412704754,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 6071
    },
    {
      "epoch": 2.425888933280064,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 6072
    },
    {
      "epoch": 2.4262884538553737,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 6073
    },
    {
      "epoch": 2.426687974430683,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 6074
    },
    {
      "epoch": 2.427087495005993,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 6075
    },
    {
      "epoch": 2.4274870155813026,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 6076
    },
    {
      "epoch": 2.427886536156612,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 6077
    },
    {
      "epoch": 2.4282860567319218,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 6078
    },
    {
      "epoch": 2.428685577307231,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 6079
    },
    {
      "epoch": 2.429085097882541,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 6080
    },
    {
      "epoch": 2.4294846184578507,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 6081
    },
    {
      "epoch": 2.42988413903316,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 6082
    },
    {
      "epoch": 2.43028365960847,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 6083
    },
    {
      "epoch": 2.4306831801837796,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 6084
    },
    {
      "epoch": 2.431082700759089,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 6085
    },
    {
      "epoch": 2.4314822213343987,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 6086
    },
    {
      "epoch": 2.4318817419097085,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 6087
    },
    {
      "epoch": 2.432281262485018,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 6088
    },
    {
      "epoch": 2.4326807830603276,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 6089
    },
    {
      "epoch": 2.433080303635637,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 6090
    },
    {
      "epoch": 2.4334798242109468,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 6091
    },
    {
      "epoch": 2.4338793447862566,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 6092
    },
    {
      "epoch": 2.4342788653615663,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 6093
    },
    {
      "epoch": 2.4346783859368757,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 6094
    },
    {
      "epoch": 2.4350779065121855,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0824,
      "step": 6095
    },
    {
      "epoch": 2.435477427087495,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 6096
    },
    {
      "epoch": 2.4358769476628046,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 6097
    },
    {
      "epoch": 2.4362764682381144,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 6098
    },
    {
      "epoch": 2.4366759888134237,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 6099
    },
    {
      "epoch": 2.4370755093887335,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 6100
    },
    {
      "epoch": 2.4374750299640433,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 6101
    },
    {
      "epoch": 2.4378745505393526,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 6102
    },
    {
      "epoch": 2.4382740711146624,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 6103
    },
    {
      "epoch": 2.438673591689972,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 6104
    },
    {
      "epoch": 2.4390731122652816,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 6105
    },
    {
      "epoch": 2.4394726328405913,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 6106
    },
    {
      "epoch": 2.4398721534159007,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 6107
    },
    {
      "epoch": 2.4402716739912105,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 6108
    },
    {
      "epoch": 2.4406711945665203,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 6109
    },
    {
      "epoch": 2.44107071514183,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 6110
    },
    {
      "epoch": 2.4414702357171394,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 6111
    },
    {
      "epoch": 2.441869756292449,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 6112
    },
    {
      "epoch": 2.4422692768677585,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 6113
    },
    {
      "epoch": 2.4426687974430683,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 6114
    },
    {
      "epoch": 2.443068318018378,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 6115
    },
    {
      "epoch": 2.4434678385936874,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 6116
    },
    {
      "epoch": 2.443867359168997,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 6117
    },
    {
      "epoch": 2.444266879744307,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 6118
    },
    {
      "epoch": 2.4446664003196164,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 6119
    },
    {
      "epoch": 2.445065920894926,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 6120
    },
    {
      "epoch": 2.445465441470236,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 6121
    },
    {
      "epoch": 2.4458649620455453,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 6122
    },
    {
      "epoch": 2.446264482620855,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 6123
    },
    {
      "epoch": 2.4466640031961644,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 6124
    },
    {
      "epoch": 2.447063523771474,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 6125
    },
    {
      "epoch": 2.447463044346784,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 6126
    },
    {
      "epoch": 2.4478625649220933,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 6127
    },
    {
      "epoch": 2.448262085497403,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 6128
    },
    {
      "epoch": 2.448661606072713,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 6129
    },
    {
      "epoch": 2.4490611266480222,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 6130
    },
    {
      "epoch": 2.449460647223332,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 6131
    },
    {
      "epoch": 2.449860167798642,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 6132
    },
    {
      "epoch": 2.450259688373951,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 6133
    },
    {
      "epoch": 2.450659208949261,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 6134
    },
    {
      "epoch": 2.4510587295245703,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 6135
    },
    {
      "epoch": 2.45145825009988,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 6136
    },
    {
      "epoch": 2.45185777067519,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 6137
    },
    {
      "epoch": 2.4522572912504996,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 6138
    },
    {
      "epoch": 2.452656811825809,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 6139
    },
    {
      "epoch": 2.4530563324011188,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 6140
    },
    {
      "epoch": 2.453455852976428,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 6141
    },
    {
      "epoch": 2.453855373551738,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 6142
    },
    {
      "epoch": 2.4542548941270477,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 6143
    },
    {
      "epoch": 2.454654414702357,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 6144
    },
    {
      "epoch": 2.455053935277667,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 6145
    },
    {
      "epoch": 2.4554534558529766,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 6146
    },
    {
      "epoch": 2.455852976428286,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 6147
    },
    {
      "epoch": 2.4562524970035957,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0665,
      "step": 6148
    },
    {
      "epoch": 2.4566520175789055,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 6149
    },
    {
      "epoch": 2.457051538154215,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 6150
    },
    {
      "epoch": 2.4574510587295246,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 6151
    },
    {
      "epoch": 2.457850579304834,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 6152
    },
    {
      "epoch": 2.4582500998801438,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 6153
    },
    {
      "epoch": 2.4586496204554535,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 6154
    },
    {
      "epoch": 2.4590491410307633,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 6155
    },
    {
      "epoch": 2.4594486616060727,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 6156
    },
    {
      "epoch": 2.4598481821813825,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 6157
    },
    {
      "epoch": 2.460247702756692,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 6158
    },
    {
      "epoch": 2.4606472233320016,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 6159
    },
    {
      "epoch": 2.4610467439073114,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 6160
    },
    {
      "epoch": 2.4614462644826207,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 6161
    },
    {
      "epoch": 2.4618457850579305,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 6162
    },
    {
      "epoch": 2.4622453056332403,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 6163
    },
    {
      "epoch": 2.4626448262085496,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 6164
    },
    {
      "epoch": 2.4630443467838594,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 6165
    },
    {
      "epoch": 2.463443867359169,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 6166
    },
    {
      "epoch": 2.4638433879344785,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 6167
    },
    {
      "epoch": 2.4642429085097883,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 6168
    },
    {
      "epoch": 2.4646424290850977,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 6169
    },
    {
      "epoch": 2.4650419496604075,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0651,
      "step": 6170
    },
    {
      "epoch": 2.4654414702357172,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 6171
    },
    {
      "epoch": 2.4658409908110266,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 6172
    },
    {
      "epoch": 2.4662405113863364,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 6173
    },
    {
      "epoch": 2.466640031961646,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 6174
    },
    {
      "epoch": 2.4670395525369555,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 6175
    },
    {
      "epoch": 2.4674390731122653,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 6176
    },
    {
      "epoch": 2.467838593687575,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 6177
    },
    {
      "epoch": 2.4682381142628844,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 6178
    },
    {
      "epoch": 2.468637634838194,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 6179
    },
    {
      "epoch": 2.4690371554135035,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 6180
    },
    {
      "epoch": 2.4694366759888133,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 6181
    },
    {
      "epoch": 2.469836196564123,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 6182
    },
    {
      "epoch": 2.470235717139433,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 6183
    },
    {
      "epoch": 2.4706352377147422,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 6184
    },
    {
      "epoch": 2.471034758290052,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 6185
    },
    {
      "epoch": 2.4714342788653614,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 6186
    },
    {
      "epoch": 2.471833799440671,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 6187
    },
    {
      "epoch": 2.472233320015981,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 6188
    },
    {
      "epoch": 2.4726328405912903,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 6189
    },
    {
      "epoch": 2.4730323611666,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 6190
    },
    {
      "epoch": 2.47343188174191,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 6191
    },
    {
      "epoch": 2.473831402317219,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 6192
    },
    {
      "epoch": 2.474230922892529,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 6193
    },
    {
      "epoch": 2.474630443467839,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 6194
    },
    {
      "epoch": 2.475029964043148,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 6195
    },
    {
      "epoch": 2.475429484618458,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 6196
    },
    {
      "epoch": 2.4758290051937673,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 6197
    },
    {
      "epoch": 2.476228525769077,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 6198
    },
    {
      "epoch": 2.476628046344387,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 6199
    },
    {
      "epoch": 2.4770275669196966,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 6200
    },
    {
      "epoch": 2.477427087495006,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 6201
    },
    {
      "epoch": 2.4778266080703157,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 6202
    },
    {
      "epoch": 2.478226128645625,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 6203
    },
    {
      "epoch": 2.478625649220935,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0651,
      "step": 6204
    },
    {
      "epoch": 2.4790251697962447,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 6205
    },
    {
      "epoch": 2.479424690371554,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 6206
    },
    {
      "epoch": 2.479824210946864,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005,
      "loss": 1.07,
      "step": 6207
    },
    {
      "epoch": 2.4802237315221736,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 6208
    },
    {
      "epoch": 2.480623252097483,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 6209
    },
    {
      "epoch": 2.4810227726727927,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 6210
    },
    {
      "epoch": 2.4814222932481025,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 6211
    },
    {
      "epoch": 2.481821813823412,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 6212
    },
    {
      "epoch": 2.4822213343987216,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 6213
    },
    {
      "epoch": 2.482620854974031,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 6214
    },
    {
      "epoch": 2.4830203755493407,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 6215
    },
    {
      "epoch": 2.4834198961246505,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 6216
    },
    {
      "epoch": 2.48381941669996,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0704,
      "step": 6217
    },
    {
      "epoch": 2.4842189372752697,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 6218
    },
    {
      "epoch": 2.4846184578505794,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 6219
    },
    {
      "epoch": 2.485017978425889,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 6220
    },
    {
      "epoch": 2.4854174990011986,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 6221
    },
    {
      "epoch": 2.4858170195765084,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 6222
    },
    {
      "epoch": 2.4862165401518177,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 6223
    },
    {
      "epoch": 2.4866160607271275,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 6224
    },
    {
      "epoch": 2.487015581302437,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 6225
    },
    {
      "epoch": 2.4874151018777466,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 6226
    },
    {
      "epoch": 2.4878146224530564,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 6227
    },
    {
      "epoch": 2.488214143028366,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0715,
      "step": 6228
    },
    {
      "epoch": 2.4886136636036755,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 6229
    },
    {
      "epoch": 2.4890131841789853,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 6230
    },
    {
      "epoch": 2.4894127047542947,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 6231
    },
    {
      "epoch": 2.4898122253296044,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 6232
    },
    {
      "epoch": 2.4902117459049142,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 6233
    },
    {
      "epoch": 2.4906112664802236,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 6234
    },
    {
      "epoch": 2.4910107870555334,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 6235
    },
    {
      "epoch": 2.491410307630843,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 6236
    },
    {
      "epoch": 2.4918098282061525,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 6237
    },
    {
      "epoch": 2.4922093487814623,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 6238
    },
    {
      "epoch": 2.492608869356772,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 6239
    },
    {
      "epoch": 2.4930083899320814,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 6240
    },
    {
      "epoch": 2.493407910507391,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 6241
    },
    {
      "epoch": 2.4938074310827005,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 6242
    },
    {
      "epoch": 2.4942069516580103,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 6243
    },
    {
      "epoch": 2.49460647223332,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 6244
    },
    {
      "epoch": 2.49500599280863,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 6245
    },
    {
      "epoch": 2.4954055133839392,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 6246
    },
    {
      "epoch": 2.495805033959249,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 6247
    },
    {
      "epoch": 2.4962045545345584,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 6248
    },
    {
      "epoch": 2.496604075109868,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 6249
    },
    {
      "epoch": 2.497003595685178,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 6250
    },
    {
      "epoch": 2.4974031162604873,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 6251
    },
    {
      "epoch": 2.497802636835797,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 6252
    },
    {
      "epoch": 2.498202157411107,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 6253
    },
    {
      "epoch": 2.498601677986416,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 6254
    },
    {
      "epoch": 2.499001198561726,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 6255
    },
    {
      "epoch": 2.4994007191370358,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 6256
    },
    {
      "epoch": 2.499800239712345,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0718,
      "step": 6257
    },
    {
      "epoch": 2.500199760287655,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 6258
    },
    {
      "epoch": 2.5005992808629642,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 6259
    },
    {
      "epoch": 2.500998801438274,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 6260
    },
    {
      "epoch": 2.501398322013584,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 6261
    },
    {
      "epoch": 2.5017978425888936,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 6262
    },
    {
      "epoch": 2.502197363164203,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 6263
    },
    {
      "epoch": 2.5025968837395127,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 6264
    },
    {
      "epoch": 2.502996404314822,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 6265
    },
    {
      "epoch": 2.503395924890132,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 6266
    },
    {
      "epoch": 2.5037954454654416,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 6267
    },
    {
      "epoch": 2.504194966040751,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 6268
    },
    {
      "epoch": 2.5045944866160608,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 6269
    },
    {
      "epoch": 2.50499400719137,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0672,
      "step": 6270
    },
    {
      "epoch": 2.50539352776668,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 6271
    },
    {
      "epoch": 2.5057930483419897,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 6272
    },
    {
      "epoch": 2.5061925689172995,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 6273
    },
    {
      "epoch": 2.506592089492609,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 6274
    },
    {
      "epoch": 2.5069916100679186,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 6275
    },
    {
      "epoch": 2.507391130643228,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 6276
    },
    {
      "epoch": 2.5077906512185377,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0725,
      "step": 6277
    },
    {
      "epoch": 2.5081901717938475,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 6278
    },
    {
      "epoch": 2.508589692369157,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 6279
    },
    {
      "epoch": 2.5089892129444666,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 6280
    },
    {
      "epoch": 2.5093887335197764,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 6281
    },
    {
      "epoch": 2.5097882540950858,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 6282
    },
    {
      "epoch": 2.5101877746703956,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 6283
    },
    {
      "epoch": 2.5105872952457053,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 6284
    },
    {
      "epoch": 2.5109868158210147,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 6285
    },
    {
      "epoch": 2.5113863363963245,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 6286
    },
    {
      "epoch": 2.511785856971634,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 6287
    },
    {
      "epoch": 2.5121853775469436,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 6288
    },
    {
      "epoch": 2.5125848981222534,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 6289
    },
    {
      "epoch": 2.512984418697563,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 6290
    },
    {
      "epoch": 2.5133839392728725,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 6291
    },
    {
      "epoch": 2.5137834598481823,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 6292
    },
    {
      "epoch": 2.5141829804234916,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 6293
    },
    {
      "epoch": 2.5145825009988014,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 6294
    },
    {
      "epoch": 2.514982021574111,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 6295
    },
    {
      "epoch": 2.5153815421494206,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0749,
      "step": 6296
    },
    {
      "epoch": 2.5157810627247303,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 6297
    },
    {
      "epoch": 2.5161805833000397,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 6298
    },
    {
      "epoch": 2.5165801038753495,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 6299
    },
    {
      "epoch": 2.5169796244506593,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 6300
    },
    {
      "epoch": 2.517379145025969,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 6301
    },
    {
      "epoch": 2.5177786656012784,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 6302
    },
    {
      "epoch": 2.518178186176588,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 6303
    },
    {
      "epoch": 2.5185777067518975,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 6304
    },
    {
      "epoch": 2.5189772273272073,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 6305
    },
    {
      "epoch": 2.519376747902517,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 6306
    },
    {
      "epoch": 2.519776268477827,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 6307
    },
    {
      "epoch": 2.520175789053136,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 6308
    },
    {
      "epoch": 2.520575309628446,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 6309
    },
    {
      "epoch": 2.5209748302037553,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 6310
    },
    {
      "epoch": 2.521374350779065,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 6311
    },
    {
      "epoch": 2.521773871354375,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 6312
    },
    {
      "epoch": 2.5221733919296843,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 6313
    },
    {
      "epoch": 2.522572912504994,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 6314
    },
    {
      "epoch": 2.5229724330803034,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 6315
    },
    {
      "epoch": 2.523371953655613,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 6316
    },
    {
      "epoch": 2.523771474230923,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 6317
    },
    {
      "epoch": 2.5241709948062327,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 6318
    },
    {
      "epoch": 2.524570515381542,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 6319
    },
    {
      "epoch": 2.524970035956852,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 6320
    },
    {
      "epoch": 2.525369556532161,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 6321
    },
    {
      "epoch": 2.525769077107471,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 6322
    },
    {
      "epoch": 2.526168597682781,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 6323
    },
    {
      "epoch": 2.52656811825809,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 6324
    },
    {
      "epoch": 2.5269676388334,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 6325
    },
    {
      "epoch": 2.5273671594087097,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 6326
    },
    {
      "epoch": 2.527766679984019,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 6327
    },
    {
      "epoch": 2.528166200559329,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 6328
    },
    {
      "epoch": 2.5285657211346386,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 6329
    },
    {
      "epoch": 2.528965241709948,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 6330
    },
    {
      "epoch": 2.5293647622852578,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 6331
    },
    {
      "epoch": 2.529764282860567,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 6332
    },
    {
      "epoch": 2.530163803435877,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 6333
    },
    {
      "epoch": 2.5305633240111867,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 6334
    },
    {
      "epoch": 2.5309628445864965,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 6335
    },
    {
      "epoch": 2.531362365161806,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 6336
    },
    {
      "epoch": 2.5317618857371156,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 6337
    },
    {
      "epoch": 2.532161406312425,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 6338
    },
    {
      "epoch": 2.5325609268877347,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 6339
    },
    {
      "epoch": 2.5329604474630445,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 6340
    },
    {
      "epoch": 2.533359968038354,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 6341
    },
    {
      "epoch": 2.5337594886136636,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 6342
    },
    {
      "epoch": 2.534159009188973,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 6343
    },
    {
      "epoch": 2.5345585297642828,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 6344
    },
    {
      "epoch": 2.5349580503395925,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 6345
    },
    {
      "epoch": 2.5353575709149023,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 6346
    },
    {
      "epoch": 2.5357570914902117,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 6347
    },
    {
      "epoch": 2.5361566120655215,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 6348
    },
    {
      "epoch": 2.536556132640831,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 6349
    },
    {
      "epoch": 2.5369556532161406,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 6350
    },
    {
      "epoch": 2.5373551737914504,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 6351
    },
    {
      "epoch": 2.53775469436676,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 6352
    },
    {
      "epoch": 2.5381542149420695,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 6353
    },
    {
      "epoch": 2.5385537355173793,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0722,
      "step": 6354
    },
    {
      "epoch": 2.5389532560926886,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 6355
    },
    {
      "epoch": 2.5393527766679984,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 6356
    },
    {
      "epoch": 2.539752297243308,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 6357
    },
    {
      "epoch": 2.5401518178186175,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 6358
    },
    {
      "epoch": 2.5405513383939273,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0722,
      "step": 6359
    },
    {
      "epoch": 2.5409508589692367,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 6360
    },
    {
      "epoch": 2.5413503795445465,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 6361
    },
    {
      "epoch": 2.5417499001198562,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.069,
      "step": 6362
    },
    {
      "epoch": 2.542149420695166,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 6363
    },
    {
      "epoch": 2.5425489412704754,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0679,
      "step": 6364
    },
    {
      "epoch": 2.542948461845785,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 6365
    },
    {
      "epoch": 2.5433479824210945,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 6366
    },
    {
      "epoch": 2.5437475029964043,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 6367
    },
    {
      "epoch": 2.544147023571714,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 6368
    },
    {
      "epoch": 2.5445465441470234,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 6369
    },
    {
      "epoch": 2.544946064722333,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 6370
    },
    {
      "epoch": 2.545345585297643,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 6371
    },
    {
      "epoch": 2.5457451058729523,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 6372
    },
    {
      "epoch": 2.546144626448262,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 6373
    },
    {
      "epoch": 2.546544147023572,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 6374
    },
    {
      "epoch": 2.5469436675988812,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 6375
    },
    {
      "epoch": 2.547343188174191,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 6376
    },
    {
      "epoch": 2.5477427087495004,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 6377
    },
    {
      "epoch": 2.54814222932481,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 6378
    },
    {
      "epoch": 2.54854174990012,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 6379
    },
    {
      "epoch": 2.5489412704754297,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 6380
    },
    {
      "epoch": 2.549340791050739,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 6381
    },
    {
      "epoch": 2.549740311626049,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 6382
    },
    {
      "epoch": 2.550139832201358,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 6383
    },
    {
      "epoch": 2.550539352776668,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 6384
    },
    {
      "epoch": 2.5509388733519778,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 6385
    },
    {
      "epoch": 2.551338393927287,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 6386
    },
    {
      "epoch": 2.551737914502597,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 6387
    },
    {
      "epoch": 2.5521374350779062,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 6388
    },
    {
      "epoch": 2.552536955653216,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 6389
    },
    {
      "epoch": 2.552936476228526,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 6390
    },
    {
      "epoch": 2.5533359968038356,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 6391
    },
    {
      "epoch": 2.553735517379145,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 6392
    },
    {
      "epoch": 2.5541350379544547,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 6393
    },
    {
      "epoch": 2.554534558529764,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 6394
    },
    {
      "epoch": 2.554934079105074,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 6395
    },
    {
      "epoch": 2.5553335996803836,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 6396
    },
    {
      "epoch": 2.5557331202556934,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 6397
    },
    {
      "epoch": 2.556132640831003,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 6398
    },
    {
      "epoch": 2.5565321614063126,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 6399
    },
    {
      "epoch": 2.556931681981622,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 6400
    },
    {
      "epoch": 2.5573312025569317,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 6401
    },
    {
      "epoch": 2.5577307231322415,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 6402
    },
    {
      "epoch": 2.558130243707551,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 6403
    },
    {
      "epoch": 2.5585297642828606,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 6404
    },
    {
      "epoch": 2.55892928485817,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 6405
    },
    {
      "epoch": 2.5593288054334797,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0732,
      "step": 6406
    },
    {
      "epoch": 2.5597283260087895,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 6407
    },
    {
      "epoch": 2.5601278465840993,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 6408
    },
    {
      "epoch": 2.5605273671594087,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 6409
    },
    {
      "epoch": 2.5609268877347184,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 6410
    },
    {
      "epoch": 2.561326408310028,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 6411
    },
    {
      "epoch": 2.5617259288853376,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0706,
      "step": 6412
    },
    {
      "epoch": 2.5621254494606474,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 6413
    },
    {
      "epoch": 2.5625249700359567,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0688,
      "step": 6414
    },
    {
      "epoch": 2.5629244906112665,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 6415
    },
    {
      "epoch": 2.5633240111865763,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 6416
    },
    {
      "epoch": 2.5637235317618856,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 6417
    },
    {
      "epoch": 2.5641230523371954,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 6418
    },
    {
      "epoch": 2.564522572912505,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 6419
    },
    {
      "epoch": 2.5649220934878145,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 6420
    },
    {
      "epoch": 2.5653216140631243,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 6421
    },
    {
      "epoch": 2.5657211346384337,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 6422
    },
    {
      "epoch": 2.5661206552137434,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 6423
    },
    {
      "epoch": 2.5665201757890532,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 6424
    },
    {
      "epoch": 2.566919696364363,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 6425
    },
    {
      "epoch": 2.5673192169396724,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 6426
    },
    {
      "epoch": 2.567718737514982,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 6427
    },
    {
      "epoch": 2.5681182580902915,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 6428
    },
    {
      "epoch": 2.5685177786656013,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 6429
    },
    {
      "epoch": 2.568917299240911,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 6430
    },
    {
      "epoch": 2.5693168198162204,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 6431
    },
    {
      "epoch": 2.56971634039153,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 6432
    },
    {
      "epoch": 2.5701158609668395,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.069,
      "step": 6433
    },
    {
      "epoch": 2.5705153815421493,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 6434
    },
    {
      "epoch": 2.570914902117459,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 6435
    },
    {
      "epoch": 2.571314422692769,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 6436
    },
    {
      "epoch": 2.5717139432680782,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 6437
    },
    {
      "epoch": 2.572113463843388,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 6438
    },
    {
      "epoch": 2.5725129844186974,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 6439
    },
    {
      "epoch": 2.572912504994007,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 6440
    },
    {
      "epoch": 2.573312025569317,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 6441
    },
    {
      "epoch": 2.5737115461446267,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.069,
      "step": 6442
    },
    {
      "epoch": 2.574111066719936,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 6443
    },
    {
      "epoch": 2.574510587295246,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 6444
    },
    {
      "epoch": 2.574910107870555,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 6445
    },
    {
      "epoch": 2.575309628445865,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 6446
    },
    {
      "epoch": 2.5757091490211748,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 6447
    },
    {
      "epoch": 2.576108669596484,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 6448
    },
    {
      "epoch": 2.576508190171794,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 6449
    },
    {
      "epoch": 2.5769077107471032,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 6450
    },
    {
      "epoch": 2.577307231322413,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 6451
    },
    {
      "epoch": 2.577706751897723,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 6452
    },
    {
      "epoch": 2.5781062724730326,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 6453
    },
    {
      "epoch": 2.578505793048342,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 6454
    },
    {
      "epoch": 2.5789053136236517,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 6455
    },
    {
      "epoch": 2.579304834198961,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 6456
    },
    {
      "epoch": 2.579704354774271,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 6457
    },
    {
      "epoch": 2.5801038753495806,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 6458
    },
    {
      "epoch": 2.58050339592489,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 6459
    },
    {
      "epoch": 2.5809029165001998,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 6460
    },
    {
      "epoch": 2.5813024370755095,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 6461
    },
    {
      "epoch": 2.581701957650819,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 6462
    },
    {
      "epoch": 2.5821014782261287,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 6463
    },
    {
      "epoch": 2.5825009988014385,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 6464
    },
    {
      "epoch": 2.582900519376748,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 6465
    },
    {
      "epoch": 2.5833000399520576,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 6466
    },
    {
      "epoch": 2.583699560527367,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 6467
    },
    {
      "epoch": 2.5840990811026767,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 6468
    },
    {
      "epoch": 2.5844986016779865,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 6469
    },
    {
      "epoch": 2.5848981222532963,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 6470
    },
    {
      "epoch": 2.5852976428286056,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 6471
    },
    {
      "epoch": 2.5856971634039154,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 6472
    },
    {
      "epoch": 2.5860966839792248,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 6473
    },
    {
      "epoch": 2.5864962045545346,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 6474
    },
    {
      "epoch": 2.5868957251298443,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 6475
    },
    {
      "epoch": 2.5872952457051537,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 6476
    },
    {
      "epoch": 2.5876947662804635,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 6477
    },
    {
      "epoch": 2.588094286855773,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0651,
      "step": 6478
    },
    {
      "epoch": 2.5884938074310826,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 6479
    },
    {
      "epoch": 2.5888933280063924,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 6480
    },
    {
      "epoch": 2.589292848581702,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 6481
    },
    {
      "epoch": 2.5896923691570115,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 6482
    },
    {
      "epoch": 2.5900918897323213,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0681,
      "step": 6483
    },
    {
      "epoch": 2.5904914103076306,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 6484
    },
    {
      "epoch": 2.5908909308829404,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 6485
    },
    {
      "epoch": 2.59129045145825,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 6486
    },
    {
      "epoch": 2.59168997203356,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 6487
    },
    {
      "epoch": 2.5920894926088693,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 6488
    },
    {
      "epoch": 2.592489013184179,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 6489
    },
    {
      "epoch": 2.5928885337594885,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 6490
    },
    {
      "epoch": 2.5932880543347983,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 6491
    },
    {
      "epoch": 2.593687574910108,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 6492
    },
    {
      "epoch": 2.5940870954854174,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0693,
      "step": 6493
    },
    {
      "epoch": 2.594486616060727,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 6494
    },
    {
      "epoch": 2.5948861366360365,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 6495
    },
    {
      "epoch": 2.5952856572113463,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 6496
    },
    {
      "epoch": 2.595685177786656,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 6497
    },
    {
      "epoch": 2.596084698361966,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 6498
    },
    {
      "epoch": 2.596484218937275,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 6499
    },
    {
      "epoch": 2.596883739512585,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 6500
    },
    {
      "epoch": 2.5972832600878943,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 6501
    },
    {
      "epoch": 2.597682780663204,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 6502
    },
    {
      "epoch": 2.598082301238514,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 6503
    },
    {
      "epoch": 2.5984818218138233,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 6504
    },
    {
      "epoch": 2.598881342389133,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 6505
    },
    {
      "epoch": 2.599280862964443,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 6506
    },
    {
      "epoch": 2.599680383539752,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 6507
    },
    {
      "epoch": 2.600079904115062,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 6508
    },
    {
      "epoch": 2.6004794246903717,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 6509
    },
    {
      "epoch": 2.600878945265681,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 6510
    },
    {
      "epoch": 2.601278465840991,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 6511
    },
    {
      "epoch": 2.6016779864163,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 6512
    },
    {
      "epoch": 2.60207750699161,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 6513
    },
    {
      "epoch": 2.60247702756692,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 6514
    },
    {
      "epoch": 2.6028765481422296,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 6515
    },
    {
      "epoch": 2.603276068717539,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 6516
    },
    {
      "epoch": 2.6036755892928487,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 6517
    },
    {
      "epoch": 2.604075109868158,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 6518
    },
    {
      "epoch": 2.604474630443468,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 6519
    },
    {
      "epoch": 2.6048741510187776,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 6520
    },
    {
      "epoch": 2.605273671594087,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 6521
    },
    {
      "epoch": 2.6056731921693967,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 6522
    },
    {
      "epoch": 2.606072712744706,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 6523
    },
    {
      "epoch": 2.606472233320016,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 6524
    },
    {
      "epoch": 2.6068717538953257,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 6525
    },
    {
      "epoch": 2.6072712744706354,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 6526
    },
    {
      "epoch": 2.607670795045945,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 6527
    },
    {
      "epoch": 2.6080703156212546,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 6528
    },
    {
      "epoch": 2.608469836196564,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 6529
    },
    {
      "epoch": 2.6088693567718737,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 6530
    },
    {
      "epoch": 2.6092688773471835,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 6531
    },
    {
      "epoch": 2.6096683979224933,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 6532
    },
    {
      "epoch": 2.6100679184978026,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 6533
    },
    {
      "epoch": 2.6104674390731124,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 6534
    },
    {
      "epoch": 2.6108669596484217,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 6535
    },
    {
      "epoch": 2.6112664802237315,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 6536
    },
    {
      "epoch": 2.6116660007990413,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 6537
    },
    {
      "epoch": 2.6120655213743507,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 6538
    },
    {
      "epoch": 2.6124650419496604,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 6539
    },
    {
      "epoch": 2.61286456252497,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 6540
    },
    {
      "epoch": 2.6132640831002796,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 6541
    },
    {
      "epoch": 2.6136636036755894,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 6542
    },
    {
      "epoch": 2.614063124250899,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 6543
    },
    {
      "epoch": 2.6144626448262085,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 6544
    },
    {
      "epoch": 2.6148621654015183,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 6545
    },
    {
      "epoch": 2.6152616859768276,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 6546
    },
    {
      "epoch": 2.6156612065521374,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 6547
    },
    {
      "epoch": 2.616060727127447,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 6548
    },
    {
      "epoch": 2.6164602477027565,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 6549
    },
    {
      "epoch": 2.6168597682780663,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 6550
    },
    {
      "epoch": 2.617259288853376,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 6551
    },
    {
      "epoch": 2.6176588094286855,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 6552
    },
    {
      "epoch": 2.6180583300039952,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 6553
    },
    {
      "epoch": 2.618457850579305,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 6554
    },
    {
      "epoch": 2.6188573711546144,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 6555
    },
    {
      "epoch": 2.619256891729924,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 6556
    },
    {
      "epoch": 2.6196564123052335,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 6557
    },
    {
      "epoch": 2.6200559328805433,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 6558
    },
    {
      "epoch": 2.620455453455853,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 6559
    },
    {
      "epoch": 2.620854974031163,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 6560
    },
    {
      "epoch": 2.621254494606472,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 6561
    },
    {
      "epoch": 2.621654015181782,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 6562
    },
    {
      "epoch": 2.6220535357570913,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 6563
    },
    {
      "epoch": 2.622453056332401,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 6564
    },
    {
      "epoch": 2.622852576907711,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 6565
    },
    {
      "epoch": 2.6232520974830202,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 6566
    },
    {
      "epoch": 2.62365161805833,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 6567
    },
    {
      "epoch": 2.6240511386336394,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 6568
    },
    {
      "epoch": 2.624450659208949,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 6569
    },
    {
      "epoch": 2.624850179784259,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 6570
    },
    {
      "epoch": 2.6252497003595687,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 6571
    },
    {
      "epoch": 2.625649220934878,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 6572
    },
    {
      "epoch": 2.626048741510188,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 6573
    },
    {
      "epoch": 2.626448262085497,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 6574
    },
    {
      "epoch": 2.626847782660807,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 6575
    },
    {
      "epoch": 2.6272473032361168,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 6576
    },
    {
      "epoch": 2.6276468238114266,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 6577
    },
    {
      "epoch": 2.628046344386736,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 6578
    },
    {
      "epoch": 2.6284458649620457,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 6579
    },
    {
      "epoch": 2.628845385537355,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 6580
    },
    {
      "epoch": 2.629244906112665,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 6581
    },
    {
      "epoch": 2.6296444266879746,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 6582
    },
    {
      "epoch": 2.630043947263284,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 6583
    },
    {
      "epoch": 2.6304434678385937,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 6584
    },
    {
      "epoch": 2.630842988413903,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 6585
    },
    {
      "epoch": 2.631242508989213,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 6586
    },
    {
      "epoch": 2.6316420295645226,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 6587
    },
    {
      "epoch": 2.6320415501398324,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 6588
    },
    {
      "epoch": 2.6324410707151418,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 6589
    },
    {
      "epoch": 2.6328405912904516,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 6590
    },
    {
      "epoch": 2.633240111865761,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 6591
    },
    {
      "epoch": 2.6336396324410707,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 6592
    },
    {
      "epoch": 2.6340391530163805,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 6593
    },
    {
      "epoch": 2.63443867359169,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 6594
    },
    {
      "epoch": 2.6348381941669996,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 6595
    },
    {
      "epoch": 2.6352377147423094,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 6596
    },
    {
      "epoch": 2.6356372353176187,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 6597
    },
    {
      "epoch": 2.6360367558929285,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 6598
    },
    {
      "epoch": 2.6364362764682383,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 6599
    },
    {
      "epoch": 2.6368357970435476,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 6600
    },
    {
      "epoch": 2.6372353176188574,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 6601
    },
    {
      "epoch": 2.6376348381941668,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 6602
    },
    {
      "epoch": 2.6380343587694766,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 6603
    },
    {
      "epoch": 2.6384338793447863,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 6604
    },
    {
      "epoch": 2.638833399920096,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 6605
    },
    {
      "epoch": 2.6392329204954055,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 6606
    },
    {
      "epoch": 2.6396324410707153,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 6607
    },
    {
      "epoch": 2.6400319616460246,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 6608
    },
    {
      "epoch": 2.6404314822213344,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 6609
    },
    {
      "epoch": 2.640831002796644,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 6610
    },
    {
      "epoch": 2.6412305233719535,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 6611
    },
    {
      "epoch": 2.6416300439472633,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 6612
    },
    {
      "epoch": 2.6420295645225726,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 6613
    },
    {
      "epoch": 2.6424290850978824,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 6614
    },
    {
      "epoch": 2.642828605673192,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 6615
    },
    {
      "epoch": 2.643228126248502,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 6616
    },
    {
      "epoch": 2.6436276468238114,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 6617
    },
    {
      "epoch": 2.644027167399121,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 6618
    },
    {
      "epoch": 2.6444266879744305,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 6619
    },
    {
      "epoch": 2.6448262085497403,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 6620
    },
    {
      "epoch": 2.64522572912505,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 6621
    },
    {
      "epoch": 2.64562524970036,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 6622
    },
    {
      "epoch": 2.646024770275669,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 6623
    },
    {
      "epoch": 2.646424290850979,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 6624
    },
    {
      "epoch": 2.6468238114262883,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 6625
    },
    {
      "epoch": 2.647223332001598,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 6626
    },
    {
      "epoch": 2.647622852576908,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 6627
    },
    {
      "epoch": 2.6480223731522172,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 6628
    },
    {
      "epoch": 2.648421893727527,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 6629
    },
    {
      "epoch": 2.6488214143028364,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 6630
    },
    {
      "epoch": 2.649220934878146,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 6631
    },
    {
      "epoch": 2.649620455453456,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 6632
    },
    {
      "epoch": 2.6500199760287657,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 6633
    },
    {
      "epoch": 2.650419496604075,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 6634
    },
    {
      "epoch": 2.650819017179385,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 6635
    },
    {
      "epoch": 2.651218537754694,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 6636
    },
    {
      "epoch": 2.651618058330004,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 6637
    },
    {
      "epoch": 2.6520175789053138,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0598,
      "step": 6638
    },
    {
      "epoch": 2.652417099480623,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 6639
    },
    {
      "epoch": 2.652816620055933,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 6640
    },
    {
      "epoch": 2.6532161406312427,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 6641
    },
    {
      "epoch": 2.653615661206552,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 6642
    },
    {
      "epoch": 2.654015181781862,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 6643
    },
    {
      "epoch": 2.6544147023571716,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 6644
    },
    {
      "epoch": 2.654814222932481,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 6645
    },
    {
      "epoch": 2.6552137435077907,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 6646
    },
    {
      "epoch": 2.6556132640831,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 6647
    },
    {
      "epoch": 2.65601278465841,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 6648
    },
    {
      "epoch": 2.6564123052337196,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 6649
    },
    {
      "epoch": 2.6568118258090294,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 6650
    },
    {
      "epoch": 2.6572113463843388,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 6651
    },
    {
      "epoch": 2.6576108669596485,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 6652
    },
    {
      "epoch": 2.658010387534958,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 6653
    },
    {
      "epoch": 2.6584099081102677,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 6654
    },
    {
      "epoch": 2.6588094286855775,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 6655
    },
    {
      "epoch": 2.659208949260887,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 6656
    },
    {
      "epoch": 2.6596084698361966,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0708,
      "step": 6657
    },
    {
      "epoch": 2.660007990411506,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 6658
    },
    {
      "epoch": 2.6604075109868157,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 6659
    },
    {
      "epoch": 2.6608070315621255,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 6660
    },
    {
      "epoch": 2.6612065521374353,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 6661
    },
    {
      "epoch": 2.6616060727127446,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 6662
    },
    {
      "epoch": 2.6620055932880544,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 6663
    },
    {
      "epoch": 2.6624051138633638,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 6664
    },
    {
      "epoch": 2.6628046344386735,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 6665
    },
    {
      "epoch": 2.6632041550139833,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0727,
      "step": 6666
    },
    {
      "epoch": 2.663603675589293,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 6667
    },
    {
      "epoch": 2.6640031961646025,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 6668
    },
    {
      "epoch": 2.6644027167399122,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 6669
    },
    {
      "epoch": 2.6648022373152216,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 6670
    },
    {
      "epoch": 2.6652017578905314,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 6671
    },
    {
      "epoch": 2.665601278465841,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 6672
    },
    {
      "epoch": 2.6660007990411505,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 6673
    },
    {
      "epoch": 2.6664003196164603,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 6674
    },
    {
      "epoch": 2.6667998401917696,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 6675
    },
    {
      "epoch": 2.6671993607670794,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 6676
    },
    {
      "epoch": 2.667598881342389,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 6677
    },
    {
      "epoch": 2.667998401917699,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 6678
    },
    {
      "epoch": 2.6683979224930083,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 6679
    },
    {
      "epoch": 2.668797443068318,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 6680
    },
    {
      "epoch": 2.6691969636436275,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 6681
    },
    {
      "epoch": 2.6695964842189372,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 6682
    },
    {
      "epoch": 2.669996004794247,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 6683
    },
    {
      "epoch": 2.6703955253695564,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 6684
    },
    {
      "epoch": 2.670795045944866,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 6685
    },
    {
      "epoch": 2.671194566520176,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 6686
    },
    {
      "epoch": 2.6715940870954853,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 6687
    },
    {
      "epoch": 2.671993607670795,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 6688
    },
    {
      "epoch": 2.672393128246105,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 6689
    },
    {
      "epoch": 2.672792648821414,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 6690
    },
    {
      "epoch": 2.673192169396724,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 6691
    },
    {
      "epoch": 2.6735916899720333,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 6692
    },
    {
      "epoch": 2.673991210547343,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 6693
    },
    {
      "epoch": 2.674390731122653,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 6694
    },
    {
      "epoch": 2.6747902516979627,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 6695
    },
    {
      "epoch": 2.675189772273272,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 6696
    },
    {
      "epoch": 2.675589292848582,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 6697
    },
    {
      "epoch": 2.675988813423891,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 6698
    },
    {
      "epoch": 2.676388333999201,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 6699
    },
    {
      "epoch": 2.6767878545745107,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0713,
      "step": 6700
    },
    {
      "epoch": 2.67718737514982,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 6701
    },
    {
      "epoch": 2.67758689572513,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 6702
    },
    {
      "epoch": 2.677986416300439,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 6703
    },
    {
      "epoch": 2.678385936875749,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 6704
    },
    {
      "epoch": 2.678785457451059,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 6705
    },
    {
      "epoch": 2.6791849780263686,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 6706
    },
    {
      "epoch": 2.679584498601678,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 6707
    },
    {
      "epoch": 2.6799840191769877,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 6708
    },
    {
      "epoch": 2.680383539752297,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 6709
    },
    {
      "epoch": 2.680783060327607,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 6710
    },
    {
      "epoch": 2.6811825809029166,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 6711
    },
    {
      "epoch": 2.6815821014782264,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 6712
    },
    {
      "epoch": 2.6819816220535357,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 6713
    },
    {
      "epoch": 2.6823811426288455,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 6714
    },
    {
      "epoch": 2.682780663204155,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 6715
    },
    {
      "epoch": 2.6831801837794647,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 6716
    },
    {
      "epoch": 2.6835797043547744,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 6717
    },
    {
      "epoch": 2.683979224930084,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 6718
    },
    {
      "epoch": 2.6843787455053936,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 6719
    },
    {
      "epoch": 2.684778266080703,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0812,
      "step": 6720
    },
    {
      "epoch": 2.6851777866560127,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 6721
    },
    {
      "epoch": 2.6855773072313225,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 6722
    },
    {
      "epoch": 2.6859768278066323,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 6723
    },
    {
      "epoch": 2.6863763483819416,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 6724
    },
    {
      "epoch": 2.6867758689572514,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 6725
    },
    {
      "epoch": 2.6871753895325607,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 6726
    },
    {
      "epoch": 2.6875749101078705,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 6727
    },
    {
      "epoch": 2.6879744306831803,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 6728
    },
    {
      "epoch": 2.6883739512584897,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 6729
    },
    {
      "epoch": 2.6887734718337994,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 6730
    },
    {
      "epoch": 2.6891729924091092,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 6731
    },
    {
      "epoch": 2.6895725129844186,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 6732
    },
    {
      "epoch": 2.6899720335597284,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 6733
    },
    {
      "epoch": 2.690371554135038,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 6734
    },
    {
      "epoch": 2.6907710747103475,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 6735
    },
    {
      "epoch": 2.6911705952856573,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 6736
    },
    {
      "epoch": 2.6915701158609666,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 6737
    },
    {
      "epoch": 2.6919696364362764,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 6738
    },
    {
      "epoch": 2.692369157011586,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 6739
    },
    {
      "epoch": 2.692768677586896,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 6740
    },
    {
      "epoch": 2.6931681981622053,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0687,
      "step": 6741
    },
    {
      "epoch": 2.693567718737515,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 6742
    },
    {
      "epoch": 2.6939672393128244,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 6743
    },
    {
      "epoch": 2.6943667598881342,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 6744
    },
    {
      "epoch": 2.694766280463444,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 6745
    },
    {
      "epoch": 2.6951658010387534,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 6746
    },
    {
      "epoch": 2.695565321614063,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 6747
    },
    {
      "epoch": 2.6959648421893725,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 6748
    },
    {
      "epoch": 2.6963643627646823,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 6749
    },
    {
      "epoch": 2.696763883339992,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 6750
    },
    {
      "epoch": 2.697163403915302,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 6751
    },
    {
      "epoch": 2.697562924490611,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 6752
    },
    {
      "epoch": 2.697962445065921,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 6753
    },
    {
      "epoch": 2.6983619656412303,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 6754
    },
    {
      "epoch": 2.69876148621654,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 6755
    },
    {
      "epoch": 2.69916100679185,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 6756
    },
    {
      "epoch": 2.6995605273671597,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 6757
    },
    {
      "epoch": 2.699960047942469,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 6758
    },
    {
      "epoch": 2.700359568517779,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 6759
    },
    {
      "epoch": 2.700759089093088,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 6760
    },
    {
      "epoch": 2.701158609668398,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 6761
    },
    {
      "epoch": 2.7015581302437077,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 6762
    },
    {
      "epoch": 2.701957650819017,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 6763
    },
    {
      "epoch": 2.702357171394327,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 6764
    },
    {
      "epoch": 2.702756691969636,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 6765
    },
    {
      "epoch": 2.703156212544946,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 6766
    },
    {
      "epoch": 2.7035557331202558,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 6767
    },
    {
      "epoch": 2.7039552536955656,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 6768
    },
    {
      "epoch": 2.704354774270875,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 6769
    },
    {
      "epoch": 2.7047542948461847,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 6770
    },
    {
      "epoch": 2.705153815421494,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 6771
    },
    {
      "epoch": 2.705553335996804,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 6772
    },
    {
      "epoch": 2.7059528565721136,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 6773
    },
    {
      "epoch": 2.706352377147423,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 6774
    },
    {
      "epoch": 2.7067518977227327,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 6775
    },
    {
      "epoch": 2.7071514182980425,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 6776
    },
    {
      "epoch": 2.707550938873352,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 6777
    },
    {
      "epoch": 2.7079504594486616,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 6778
    },
    {
      "epoch": 2.7083499800239714,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 6779
    },
    {
      "epoch": 2.7087495005992808,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 6780
    },
    {
      "epoch": 2.7091490211745906,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 6781
    },
    {
      "epoch": 2.7095485417499,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 6782
    },
    {
      "epoch": 2.7099480623252097,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 6783
    },
    {
      "epoch": 2.7103475829005195,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 6784
    },
    {
      "epoch": 2.7107471034758293,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 6785
    },
    {
      "epoch": 2.7111466240511386,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 6786
    },
    {
      "epoch": 2.7115461446264484,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 6787
    },
    {
      "epoch": 2.7119456652017577,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 6788
    },
    {
      "epoch": 2.7123451857770675,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 6789
    },
    {
      "epoch": 2.7127447063523773,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 6790
    },
    {
      "epoch": 2.7131442269276866,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 6791
    },
    {
      "epoch": 2.7135437475029964,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 6792
    },
    {
      "epoch": 2.7139432680783058,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 6793
    },
    {
      "epoch": 2.7143427886536156,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 6794
    },
    {
      "epoch": 2.7147423092289253,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 6795
    },
    {
      "epoch": 2.715141829804235,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 6796
    },
    {
      "epoch": 2.7155413503795445,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 6797
    },
    {
      "epoch": 2.7159408709548543,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 6798
    },
    {
      "epoch": 2.7163403915301636,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 6799
    },
    {
      "epoch": 2.7167399121054734,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 6800
    },
    {
      "epoch": 2.717139432680783,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 6801
    },
    {
      "epoch": 2.717538953256093,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 6802
    },
    {
      "epoch": 2.7179384738314023,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 6803
    },
    {
      "epoch": 2.718337994406712,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 6804
    },
    {
      "epoch": 2.7187375149820214,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 6805
    },
    {
      "epoch": 2.719137035557331,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 6806
    },
    {
      "epoch": 2.719536556132641,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 6807
    },
    {
      "epoch": 2.7199360767079503,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 6808
    },
    {
      "epoch": 2.72033559728326,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 6809
    },
    {
      "epoch": 2.7207351178585695,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 6810
    },
    {
      "epoch": 2.7211346384338793,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 6811
    },
    {
      "epoch": 2.721534159009189,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 6812
    },
    {
      "epoch": 2.721933679584499,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 6813
    },
    {
      "epoch": 2.722333200159808,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 6814
    },
    {
      "epoch": 2.722732720735118,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 6815
    },
    {
      "epoch": 2.7231322413104273,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 6816
    },
    {
      "epoch": 2.723531761885737,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 6817
    },
    {
      "epoch": 2.723931282461047,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 6818
    },
    {
      "epoch": 2.724330803036356,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 6819
    },
    {
      "epoch": 2.724730323611666,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 6820
    },
    {
      "epoch": 2.725129844186976,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 6821
    },
    {
      "epoch": 2.725529364762285,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 6822
    },
    {
      "epoch": 2.725928885337595,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 6823
    },
    {
      "epoch": 2.7263284059129047,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 6824
    },
    {
      "epoch": 2.726727926488214,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 6825
    },
    {
      "epoch": 2.727127447063524,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 6826
    },
    {
      "epoch": 2.727526967638833,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 6827
    },
    {
      "epoch": 2.727926488214143,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 6828
    },
    {
      "epoch": 2.7283260087894528,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 6829
    },
    {
      "epoch": 2.7287255293647625,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 6830
    },
    {
      "epoch": 2.729125049940072,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 6831
    },
    {
      "epoch": 2.7295245705153817,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 6832
    },
    {
      "epoch": 2.729924091090691,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 6833
    },
    {
      "epoch": 2.730323611666001,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 6834
    },
    {
      "epoch": 2.7307231322413106,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 6835
    },
    {
      "epoch": 2.73112265281662,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 6836
    },
    {
      "epoch": 2.7315221733919297,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 6837
    },
    {
      "epoch": 2.731921693967239,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 6838
    },
    {
      "epoch": 2.732321214542549,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 6839
    },
    {
      "epoch": 2.7327207351178586,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 6840
    },
    {
      "epoch": 2.7331202556931684,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 6841
    },
    {
      "epoch": 2.7335197762684778,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 6842
    },
    {
      "epoch": 2.7339192968437875,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 6843
    },
    {
      "epoch": 2.734318817419097,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 6844
    },
    {
      "epoch": 2.7347183379944067,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 6845
    },
    {
      "epoch": 2.7351178585697165,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 6846
    },
    {
      "epoch": 2.7355173791450262,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 6847
    },
    {
      "epoch": 2.7359168997203356,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 6848
    },
    {
      "epoch": 2.7363164202956454,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 6849
    },
    {
      "epoch": 2.7367159408709547,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 6850
    },
    {
      "epoch": 2.7371154614462645,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 6851
    },
    {
      "epoch": 2.7375149820215743,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 6852
    },
    {
      "epoch": 2.7379145025968836,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 6853
    },
    {
      "epoch": 2.7383140231721934,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 6854
    },
    {
      "epoch": 2.7387135437475028,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 6855
    },
    {
      "epoch": 2.7391130643228125,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 6856
    },
    {
      "epoch": 2.7395125848981223,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 6857
    },
    {
      "epoch": 2.739912105473432,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 6858
    },
    {
      "epoch": 2.7403116260487415,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 6859
    },
    {
      "epoch": 2.7407111466240512,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 6860
    },
    {
      "epoch": 2.7411106671993606,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 6861
    },
    {
      "epoch": 2.7415101877746704,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 6862
    },
    {
      "epoch": 2.74190970834998,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 6863
    },
    {
      "epoch": 2.7423092289252895,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 6864
    },
    {
      "epoch": 2.7427087495005993,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 6865
    },
    {
      "epoch": 2.743108270075909,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 6866
    },
    {
      "epoch": 2.7435077906512184,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 6867
    },
    {
      "epoch": 2.743907311226528,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 6868
    },
    {
      "epoch": 2.744306831801838,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 6869
    },
    {
      "epoch": 2.7447063523771473,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 6870
    },
    {
      "epoch": 2.745105872952457,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 6871
    },
    {
      "epoch": 2.7455053935277665,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 6872
    },
    {
      "epoch": 2.7459049141030762,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 6873
    },
    {
      "epoch": 2.746304434678386,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0717,
      "step": 6874
    },
    {
      "epoch": 2.746703955253696,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 6875
    },
    {
      "epoch": 2.747103475829005,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 6876
    },
    {
      "epoch": 2.747502996404315,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 6877
    },
    {
      "epoch": 2.7479025169796243,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 6878
    },
    {
      "epoch": 2.748302037554934,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 6879
    },
    {
      "epoch": 2.748701558130244,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 6880
    },
    {
      "epoch": 2.749101078705553,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 6881
    },
    {
      "epoch": 2.749500599280863,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 6882
    },
    {
      "epoch": 2.7499001198561723,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 6883
    },
    {
      "epoch": 2.750299640431482,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 6884
    },
    {
      "epoch": 2.750699161006792,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 6885
    },
    {
      "epoch": 2.7510986815821017,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 6886
    },
    {
      "epoch": 2.751498202157411,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 6887
    },
    {
      "epoch": 2.751897722732721,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 6888
    },
    {
      "epoch": 2.75229724330803,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0696,
      "step": 6889
    },
    {
      "epoch": 2.75269676388334,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 6890
    },
    {
      "epoch": 2.7530962844586497,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 6891
    },
    {
      "epoch": 2.7534958050339595,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 6892
    },
    {
      "epoch": 2.753895325609269,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 6893
    },
    {
      "epoch": 2.7542948461845786,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 6894
    },
    {
      "epoch": 2.754694366759888,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 6895
    },
    {
      "epoch": 2.7550938873351978,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 6896
    },
    {
      "epoch": 2.7554934079105076,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 6897
    },
    {
      "epoch": 2.755892928485817,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 6898
    },
    {
      "epoch": 2.7562924490611267,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 6899
    },
    {
      "epoch": 2.756691969636436,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 6900
    },
    {
      "epoch": 2.757091490211746,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 6901
    },
    {
      "epoch": 2.7574910107870556,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 6902
    },
    {
      "epoch": 2.7578905313623654,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 6903
    },
    {
      "epoch": 2.7582900519376747,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 6904
    },
    {
      "epoch": 2.7586895725129845,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 6905
    },
    {
      "epoch": 2.759089093088294,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 6906
    },
    {
      "epoch": 2.7594886136636037,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 6907
    },
    {
      "epoch": 2.7598881342389134,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 6908
    },
    {
      "epoch": 2.760287654814223,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 6909
    },
    {
      "epoch": 2.7606871753895326,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 6910
    },
    {
      "epoch": 2.7610866959648424,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 6911
    },
    {
      "epoch": 2.7614862165401517,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 6912
    },
    {
      "epoch": 2.7618857371154615,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 6913
    },
    {
      "epoch": 2.7622852576907713,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 6914
    },
    {
      "epoch": 2.7626847782660806,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 6915
    },
    {
      "epoch": 2.7630842988413904,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 6916
    },
    {
      "epoch": 2.7634838194166997,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0699,
      "step": 6917
    },
    {
      "epoch": 2.7638833399920095,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 6918
    },
    {
      "epoch": 2.7642828605673193,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 6919
    },
    {
      "epoch": 2.764682381142629,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0693,
      "step": 6920
    },
    {
      "epoch": 2.7650819017179384,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 6921
    },
    {
      "epoch": 2.7654814222932482,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 6922
    },
    {
      "epoch": 2.7658809428685576,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 6923
    },
    {
      "epoch": 2.7662804634438674,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 6924
    },
    {
      "epoch": 2.766679984019177,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 6925
    },
    {
      "epoch": 2.7670795045944865,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 6926
    },
    {
      "epoch": 2.7674790251697963,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 6927
    },
    {
      "epoch": 2.7678785457451056,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 6928
    },
    {
      "epoch": 2.7682780663204154,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 6929
    },
    {
      "epoch": 2.768677586895725,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 6930
    },
    {
      "epoch": 2.769077107471035,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 6931
    },
    {
      "epoch": 2.7694766280463443,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 6932
    },
    {
      "epoch": 2.769876148621654,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 6933
    },
    {
      "epoch": 2.7702756691969634,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 6934
    },
    {
      "epoch": 2.7706751897722732,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 6935
    },
    {
      "epoch": 2.771074710347583,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 6936
    },
    {
      "epoch": 2.771474230922893,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 6937
    },
    {
      "epoch": 2.771873751498202,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 6938
    },
    {
      "epoch": 2.772273272073512,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 6939
    },
    {
      "epoch": 2.7726727926488213,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 6940
    },
    {
      "epoch": 2.773072313224131,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 6941
    },
    {
      "epoch": 2.773471833799441,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 6942
    },
    {
      "epoch": 2.77387135437475,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 6943
    },
    {
      "epoch": 2.77427087495006,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 6944
    },
    {
      "epoch": 2.7746703955253693,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 6945
    },
    {
      "epoch": 2.775069916100679,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 6946
    },
    {
      "epoch": 2.775469436675989,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 6947
    },
    {
      "epoch": 2.7758689572512987,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 6948
    },
    {
      "epoch": 2.776268477826608,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 6949
    },
    {
      "epoch": 2.776667998401918,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 6950
    },
    {
      "epoch": 2.777067518977227,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 6951
    },
    {
      "epoch": 2.777467039552537,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 6952
    },
    {
      "epoch": 2.7778665601278467,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 6953
    },
    {
      "epoch": 2.778266080703156,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 6954
    },
    {
      "epoch": 2.778665601278466,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 6955
    },
    {
      "epoch": 2.7790651218537756,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 6956
    },
    {
      "epoch": 2.779464642429085,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 6957
    },
    {
      "epoch": 2.7798641630043948,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 6958
    },
    {
      "epoch": 2.7802636835797045,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 6959
    },
    {
      "epoch": 2.780663204155014,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 6960
    },
    {
      "epoch": 2.7810627247303237,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 6961
    },
    {
      "epoch": 2.781462245305633,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 6962
    },
    {
      "epoch": 2.781861765880943,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 6963
    },
    {
      "epoch": 2.7822612864562526,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 6964
    },
    {
      "epoch": 2.7826608070315624,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 6965
    },
    {
      "epoch": 2.7830603276068717,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 6966
    },
    {
      "epoch": 2.7834598481821815,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 6967
    },
    {
      "epoch": 2.783859368757491,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 6968
    },
    {
      "epoch": 2.7842588893328006,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 6969
    },
    {
      "epoch": 2.7846584099081104,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 6970
    },
    {
      "epoch": 2.7850579304834198,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 6971
    },
    {
      "epoch": 2.7854574510587295,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 6972
    },
    {
      "epoch": 2.785856971634039,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 6973
    },
    {
      "epoch": 2.7862564922093487,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 6974
    },
    {
      "epoch": 2.7866560127846585,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 6975
    },
    {
      "epoch": 2.7870555333599683,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 6976
    },
    {
      "epoch": 2.7874550539352776,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 6977
    },
    {
      "epoch": 2.7878545745105874,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 6978
    },
    {
      "epoch": 2.7882540950858967,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 6979
    },
    {
      "epoch": 2.7886536156612065,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 6980
    },
    {
      "epoch": 2.7890531362365163,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 6981
    },
    {
      "epoch": 2.789452656811826,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 6982
    },
    {
      "epoch": 2.7898521773871354,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 6983
    },
    {
      "epoch": 2.790251697962445,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 6984
    },
    {
      "epoch": 2.7906512185377546,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 6985
    },
    {
      "epoch": 2.7910507391130643,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 6986
    },
    {
      "epoch": 2.791450259688374,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 6987
    },
    {
      "epoch": 2.7918497802636835,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 6988
    },
    {
      "epoch": 2.7922493008389933,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 6989
    },
    {
      "epoch": 2.7926488214143026,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 6990
    },
    {
      "epoch": 2.7930483419896124,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 6991
    },
    {
      "epoch": 2.793447862564922,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 6992
    },
    {
      "epoch": 2.793847383140232,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 6993
    },
    {
      "epoch": 2.7942469037155413,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0737,
      "step": 6994
    },
    {
      "epoch": 2.794646424290851,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0734,
      "step": 6995
    },
    {
      "epoch": 2.7950459448661604,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 6996
    },
    {
      "epoch": 2.79544546544147,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 6997
    },
    {
      "epoch": 2.79584498601678,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 6998
    },
    {
      "epoch": 2.7962445065920893,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 6999
    },
    {
      "epoch": 2.796644027167399,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 7000
    },
    {
      "epoch": 2.797043547742709,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 7001
    },
    {
      "epoch": 2.7974430683180183,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 7002
    },
    {
      "epoch": 2.797842588893328,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 7003
    },
    {
      "epoch": 2.798242109468638,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 7004
    },
    {
      "epoch": 2.798641630043947,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 7005
    },
    {
      "epoch": 2.799041150619257,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 7006
    },
    {
      "epoch": 2.7994406711945663,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 7007
    },
    {
      "epoch": 2.799840191769876,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 7008
    },
    {
      "epoch": 2.800239712345186,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 7009
    },
    {
      "epoch": 2.8006392329204957,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 7010
    },
    {
      "epoch": 2.801038753495805,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 7011
    },
    {
      "epoch": 2.801438274071115,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 7012
    },
    {
      "epoch": 2.801837794646424,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 7013
    },
    {
      "epoch": 2.802237315221734,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 7014
    },
    {
      "epoch": 2.8026368357970437,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 7015
    },
    {
      "epoch": 2.803036356372353,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 7016
    },
    {
      "epoch": 2.803435876947663,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 7017
    },
    {
      "epoch": 2.803835397522972,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 7018
    },
    {
      "epoch": 2.804234918098282,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 7019
    },
    {
      "epoch": 2.8046344386735917,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 7020
    },
    {
      "epoch": 2.8050339592489015,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 7021
    },
    {
      "epoch": 2.805433479824211,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 7022
    },
    {
      "epoch": 2.8058330003995207,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 7023
    },
    {
      "epoch": 2.80623252097483,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0763,
      "step": 7024
    },
    {
      "epoch": 2.80663204155014,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 7025
    },
    {
      "epoch": 2.8070315621254496,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 7026
    },
    {
      "epoch": 2.8074310827007594,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 7027
    },
    {
      "epoch": 2.8078306032760687,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 7028
    },
    {
      "epoch": 2.8082301238513785,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 7029
    },
    {
      "epoch": 2.808629644426688,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 7030
    },
    {
      "epoch": 2.8090291650019976,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 7031
    },
    {
      "epoch": 2.8094286855773074,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 7032
    },
    {
      "epoch": 2.8098282061526167,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0754,
      "step": 7033
    },
    {
      "epoch": 2.8102277267279265,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 7034
    },
    {
      "epoch": 2.810627247303236,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 7035
    },
    {
      "epoch": 2.8110267678785457,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 7036
    },
    {
      "epoch": 2.8114262884538554,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 7037
    },
    {
      "epoch": 2.8118258090291652,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 7038
    },
    {
      "epoch": 2.8122253296044746,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 7039
    },
    {
      "epoch": 2.8126248501797844,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 7040
    },
    {
      "epoch": 2.8130243707550937,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 7041
    },
    {
      "epoch": 2.8134238913304035,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 7042
    },
    {
      "epoch": 2.8138234119057133,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 7043
    },
    {
      "epoch": 2.8142229324810226,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 7044
    },
    {
      "epoch": 2.8146224530563324,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 7045
    },
    {
      "epoch": 2.815021973631642,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 7046
    },
    {
      "epoch": 2.8154214942069515,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 7047
    },
    {
      "epoch": 2.8158210147822613,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 7048
    },
    {
      "epoch": 2.816220535357571,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 7049
    },
    {
      "epoch": 2.8166200559328805,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 7050
    },
    {
      "epoch": 2.8170195765081902,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 7051
    },
    {
      "epoch": 2.8174190970834996,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 7052
    },
    {
      "epoch": 2.8178186176588094,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0696,
      "step": 7053
    },
    {
      "epoch": 2.818218138234119,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 7054
    },
    {
      "epoch": 2.818617658809429,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 7055
    },
    {
      "epoch": 2.8190171793847383,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 7056
    },
    {
      "epoch": 2.819416699960048,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 7057
    },
    {
      "epoch": 2.8198162205353574,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 7058
    },
    {
      "epoch": 2.820215741110667,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 7059
    },
    {
      "epoch": 2.820615261685977,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 7060
    },
    {
      "epoch": 2.8210147822612863,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 7061
    },
    {
      "epoch": 2.821414302836596,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 7062
    },
    {
      "epoch": 2.8218138234119055,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 7063
    },
    {
      "epoch": 2.8222133439872152,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0666,
      "step": 7064
    },
    {
      "epoch": 2.822612864562525,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 7065
    },
    {
      "epoch": 2.823012385137835,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 7066
    },
    {
      "epoch": 2.823411905713144,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 7067
    },
    {
      "epoch": 2.823811426288454,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 7068
    },
    {
      "epoch": 2.8242109468637633,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 7069
    },
    {
      "epoch": 2.824610467439073,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 7070
    },
    {
      "epoch": 2.825009988014383,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 7071
    },
    {
      "epoch": 2.8254095085896926,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 7072
    },
    {
      "epoch": 2.825809029165002,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 7073
    },
    {
      "epoch": 2.8262085497403118,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0672,
      "step": 7074
    },
    {
      "epoch": 2.826608070315621,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 7075
    },
    {
      "epoch": 2.827007590890931,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 7076
    },
    {
      "epoch": 2.8274071114662407,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 7077
    },
    {
      "epoch": 2.82780663204155,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 7078
    },
    {
      "epoch": 2.82820615261686,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 7079
    },
    {
      "epoch": 2.828605673192169,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 7080
    },
    {
      "epoch": 2.829005193767479,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 7081
    },
    {
      "epoch": 2.8294047143427887,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0732,
      "step": 7082
    },
    {
      "epoch": 2.8298042349180985,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 7083
    },
    {
      "epoch": 2.830203755493408,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 7084
    },
    {
      "epoch": 2.8306032760687176,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 7085
    },
    {
      "epoch": 2.831002796644027,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 7086
    },
    {
      "epoch": 2.8314023172193368,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 7087
    },
    {
      "epoch": 2.8318018377946466,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 7088
    },
    {
      "epoch": 2.832201358369956,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 7089
    },
    {
      "epoch": 2.8326008789452657,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 7090
    },
    {
      "epoch": 2.8330003995205755,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 7091
    },
    {
      "epoch": 2.833399920095885,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 7092
    },
    {
      "epoch": 2.8337994406711946,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 7093
    },
    {
      "epoch": 2.8341989612465044,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 7094
    },
    {
      "epoch": 2.8345984818218137,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 7095
    },
    {
      "epoch": 2.8349980023971235,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 7096
    },
    {
      "epoch": 2.835397522972433,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 7097
    },
    {
      "epoch": 2.8357970435477426,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 7098
    },
    {
      "epoch": 2.8361965641230524,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 7099
    },
    {
      "epoch": 2.836596084698362,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 7100
    },
    {
      "epoch": 2.8369956052736716,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 7101
    },
    {
      "epoch": 2.8373951258489813,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 7102
    },
    {
      "epoch": 2.8377946464242907,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 7103
    },
    {
      "epoch": 2.8381941669996005,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 7104
    },
    {
      "epoch": 2.8385936875749103,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 7105
    },
    {
      "epoch": 2.8389932081502196,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 7106
    },
    {
      "epoch": 2.8393927287255294,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 7107
    },
    {
      "epoch": 2.8397922493008387,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 7108
    },
    {
      "epoch": 2.8401917698761485,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 7109
    },
    {
      "epoch": 2.8405912904514583,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 7110
    },
    {
      "epoch": 2.840990811026768,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 7111
    },
    {
      "epoch": 2.8413903316020774,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 7112
    },
    {
      "epoch": 2.841789852177387,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 7113
    },
    {
      "epoch": 2.8421893727526966,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 7114
    },
    {
      "epoch": 2.8425888933280063,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 7115
    },
    {
      "epoch": 2.842988413903316,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 7116
    },
    {
      "epoch": 2.843387934478626,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 7117
    },
    {
      "epoch": 2.8437874550539353,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 7118
    },
    {
      "epoch": 2.844186975629245,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 7119
    },
    {
      "epoch": 2.8445864962045544,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 7120
    },
    {
      "epoch": 2.844986016779864,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 7121
    },
    {
      "epoch": 2.845385537355174,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 7122
    },
    {
      "epoch": 2.8457850579304833,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 7123
    },
    {
      "epoch": 2.846184578505793,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 7124
    },
    {
      "epoch": 2.8465840990811024,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 7125
    },
    {
      "epoch": 2.8469836196564122,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 7126
    },
    {
      "epoch": 2.847383140231722,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 7127
    },
    {
      "epoch": 2.847782660807032,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 7128
    },
    {
      "epoch": 2.848182181382341,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 7129
    },
    {
      "epoch": 2.848581701957651,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 7130
    },
    {
      "epoch": 2.8489812225329603,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 7131
    },
    {
      "epoch": 2.84938074310827,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 7132
    },
    {
      "epoch": 2.84978026368358,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 7133
    },
    {
      "epoch": 2.850179784258889,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 7134
    },
    {
      "epoch": 2.850579304834199,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 7135
    },
    {
      "epoch": 2.8509788254095088,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 7136
    },
    {
      "epoch": 2.851378345984818,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 7137
    },
    {
      "epoch": 2.851777866560128,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 7138
    },
    {
      "epoch": 2.8521773871354377,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 7139
    },
    {
      "epoch": 2.852576907710747,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 7140
    },
    {
      "epoch": 2.852976428286057,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 7141
    },
    {
      "epoch": 2.853375948861366,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 7142
    },
    {
      "epoch": 2.853775469436676,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 7143
    },
    {
      "epoch": 2.8541749900119857,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 7144
    },
    {
      "epoch": 2.8545745105872955,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 7145
    },
    {
      "epoch": 2.854974031162605,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 7146
    },
    {
      "epoch": 2.8553735517379146,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0638,
      "step": 7147
    },
    {
      "epoch": 2.855773072313224,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 7148
    },
    {
      "epoch": 2.8561725928885338,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 7149
    },
    {
      "epoch": 2.8565721134638435,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 7150
    },
    {
      "epoch": 2.856971634039153,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 7151
    },
    {
      "epoch": 2.8573711546144627,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 7152
    },
    {
      "epoch": 2.857770675189772,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 7153
    },
    {
      "epoch": 2.858170195765082,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 7154
    },
    {
      "epoch": 2.8585697163403916,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 7155
    },
    {
      "epoch": 2.8589692369157014,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 7156
    },
    {
      "epoch": 2.8593687574910107,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 7157
    },
    {
      "epoch": 2.8597682780663205,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 7158
    },
    {
      "epoch": 2.86016779864163,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 7159
    },
    {
      "epoch": 2.8605673192169396,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 7160
    },
    {
      "epoch": 2.8609668397922494,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 7161
    },
    {
      "epoch": 2.861366360367559,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 7162
    },
    {
      "epoch": 2.8617658809428685,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 7163
    },
    {
      "epoch": 2.8621654015181783,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 7164
    },
    {
      "epoch": 2.8625649220934877,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 7165
    },
    {
      "epoch": 2.8629644426687975,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0693,
      "step": 7166
    },
    {
      "epoch": 2.8633639632441072,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 7167
    },
    {
      "epoch": 2.8637634838194166,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 7168
    },
    {
      "epoch": 2.8641630043947264,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 7169
    },
    {
      "epoch": 2.8645625249700357,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 7170
    },
    {
      "epoch": 2.8649620455453455,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 7171
    },
    {
      "epoch": 2.8653615661206553,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 7172
    },
    {
      "epoch": 2.865761086695965,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 7173
    },
    {
      "epoch": 2.8661606072712744,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 7174
    },
    {
      "epoch": 2.866560127846584,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 7175
    },
    {
      "epoch": 2.8669596484218935,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 7176
    },
    {
      "epoch": 2.8673591689972033,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 7177
    },
    {
      "epoch": 2.867758689572513,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 7178
    },
    {
      "epoch": 2.8681582101478225,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 7179
    },
    {
      "epoch": 2.8685577307231322,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 7180
    },
    {
      "epoch": 2.868957251298442,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 7181
    },
    {
      "epoch": 2.8693567718737514,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 7182
    },
    {
      "epoch": 2.869756292449061,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 7183
    },
    {
      "epoch": 2.870155813024371,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 7184
    },
    {
      "epoch": 2.8705553335996803,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 7185
    },
    {
      "epoch": 2.87095485417499,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 7186
    },
    {
      "epoch": 2.8713543747502994,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 7187
    },
    {
      "epoch": 2.871753895325609,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0706,
      "step": 7188
    },
    {
      "epoch": 2.872153415900919,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 7189
    },
    {
      "epoch": 2.872552936476229,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 7190
    },
    {
      "epoch": 2.872952457051538,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 7191
    },
    {
      "epoch": 2.873351977626848,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 7192
    },
    {
      "epoch": 2.8737514982021573,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 7193
    },
    {
      "epoch": 2.874151018777467,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 7194
    },
    {
      "epoch": 2.874550539352777,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 7195
    },
    {
      "epoch": 2.874950059928086,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 7196
    },
    {
      "epoch": 2.875349580503396,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 7197
    },
    {
      "epoch": 2.8757491010787053,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 7198
    },
    {
      "epoch": 2.876148621654015,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 7199
    },
    {
      "epoch": 2.876548142229325,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 7200
    },
    {
      "epoch": 2.8769476628046347,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 7201
    },
    {
      "epoch": 2.877347183379944,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 7202
    },
    {
      "epoch": 2.877746703955254,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 7203
    },
    {
      "epoch": 2.878146224530563,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 7204
    },
    {
      "epoch": 2.878545745105873,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 7205
    },
    {
      "epoch": 2.8789452656811827,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 7206
    },
    {
      "epoch": 2.8793447862564925,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 7207
    },
    {
      "epoch": 2.879744306831802,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0679,
      "step": 7208
    },
    {
      "epoch": 2.8801438274071116,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 7209
    },
    {
      "epoch": 2.880543347982421,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 7210
    },
    {
      "epoch": 2.8809428685577307,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 7211
    },
    {
      "epoch": 2.8813423891330405,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 7212
    },
    {
      "epoch": 2.88174190970835,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 7213
    },
    {
      "epoch": 2.8821414302836597,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 7214
    },
    {
      "epoch": 2.882540950858969,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 7215
    },
    {
      "epoch": 2.882940471434279,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 7216
    },
    {
      "epoch": 2.8833399920095886,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 7217
    },
    {
      "epoch": 2.8837395125848984,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 7218
    },
    {
      "epoch": 2.8841390331602077,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 7219
    },
    {
      "epoch": 2.8845385537355175,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 7220
    },
    {
      "epoch": 2.884938074310827,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 7221
    },
    {
      "epoch": 2.8853375948861366,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 7222
    },
    {
      "epoch": 2.8857371154614464,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 7223
    },
    {
      "epoch": 2.8861366360367557,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 7224
    },
    {
      "epoch": 2.8865361566120655,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 7225
    },
    {
      "epoch": 2.8869356771873753,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 7226
    },
    {
      "epoch": 2.8873351977626847,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 7227
    },
    {
      "epoch": 2.8877347183379944,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 7228
    },
    {
      "epoch": 2.8881342389133042,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 7229
    },
    {
      "epoch": 2.8885337594886136,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0598,
      "step": 7230
    },
    {
      "epoch": 2.8889332800639234,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 7231
    },
    {
      "epoch": 2.8893328006392327,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 7232
    },
    {
      "epoch": 2.8897323212145425,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0704,
      "step": 7233
    },
    {
      "epoch": 2.8901318417898523,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 7234
    },
    {
      "epoch": 2.890531362365162,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 7235
    },
    {
      "epoch": 2.8909308829404714,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 7236
    },
    {
      "epoch": 2.891330403515781,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 7237
    },
    {
      "epoch": 2.8917299240910905,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 7238
    },
    {
      "epoch": 2.8921294446664003,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 7239
    },
    {
      "epoch": 2.89252896524171,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 7240
    },
    {
      "epoch": 2.8929284858170194,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 7241
    },
    {
      "epoch": 2.8933280063923292,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 7242
    },
    {
      "epoch": 2.8937275269676386,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 7243
    },
    {
      "epoch": 2.8941270475429484,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 7244
    },
    {
      "epoch": 2.894526568118258,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 7245
    },
    {
      "epoch": 2.894926088693568,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 7246
    },
    {
      "epoch": 2.8953256092688773,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 7247
    },
    {
      "epoch": 2.895725129844187,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 7248
    },
    {
      "epoch": 2.8961246504194964,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 7249
    },
    {
      "epoch": 2.896524170994806,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 7250
    },
    {
      "epoch": 2.896923691570116,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 7251
    },
    {
      "epoch": 2.8973232121454258,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 7252
    },
    {
      "epoch": 2.897722732720735,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 7253
    },
    {
      "epoch": 2.898122253296045,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 7254
    },
    {
      "epoch": 2.8985217738713542,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 7255
    },
    {
      "epoch": 2.898921294446664,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 7256
    },
    {
      "epoch": 2.899320815021974,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 7257
    },
    {
      "epoch": 2.899720335597283,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 7258
    },
    {
      "epoch": 2.900119856172593,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 7259
    },
    {
      "epoch": 2.9005193767479023,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 7260
    },
    {
      "epoch": 2.900918897323212,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 7261
    },
    {
      "epoch": 2.901318417898522,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 7262
    },
    {
      "epoch": 2.9017179384738316,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 7263
    },
    {
      "epoch": 2.902117459049141,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 7264
    },
    {
      "epoch": 2.9025169796244508,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 7265
    },
    {
      "epoch": 2.90291650019976,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 7266
    },
    {
      "epoch": 2.90331602077507,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 7267
    },
    {
      "epoch": 2.9037155413503797,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 7268
    },
    {
      "epoch": 2.904115061925689,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 7269
    },
    {
      "epoch": 2.904514582500999,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 7270
    },
    {
      "epoch": 2.9049141030763086,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 7271
    },
    {
      "epoch": 2.905313623651618,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 7272
    },
    {
      "epoch": 2.9057131442269277,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 7273
    },
    {
      "epoch": 2.9061126648022375,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 7274
    },
    {
      "epoch": 2.906512185377547,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 7275
    },
    {
      "epoch": 2.9069117059528566,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 7276
    },
    {
      "epoch": 2.907311226528166,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 7277
    },
    {
      "epoch": 2.9077107471034758,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 7278
    },
    {
      "epoch": 2.9081102676787856,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 7279
    },
    {
      "epoch": 2.9085097882540953,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 7280
    },
    {
      "epoch": 2.9089093088294047,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 7281
    },
    {
      "epoch": 2.9093088294047145,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 7282
    },
    {
      "epoch": 2.909708349980024,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 7283
    },
    {
      "epoch": 2.9101078705553336,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 7284
    },
    {
      "epoch": 2.9105073911306434,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 7285
    },
    {
      "epoch": 2.9109069117059527,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 7286
    },
    {
      "epoch": 2.9113064322812625,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 7287
    },
    {
      "epoch": 2.911705952856572,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 7288
    },
    {
      "epoch": 2.9121054734318816,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 7289
    },
    {
      "epoch": 2.9125049940071914,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 7290
    },
    {
      "epoch": 2.912904514582501,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 7291
    },
    {
      "epoch": 2.9133040351578106,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 7292
    },
    {
      "epoch": 2.9137035557331203,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 7293
    },
    {
      "epoch": 2.9141030763084297,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 7294
    },
    {
      "epoch": 2.9145025968837395,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 7295
    },
    {
      "epoch": 2.9149021174590493,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 7296
    },
    {
      "epoch": 2.915301638034359,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 7297
    },
    {
      "epoch": 2.9157011586096684,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 7298
    },
    {
      "epoch": 2.916100679184978,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 7299
    },
    {
      "epoch": 2.9165001997602875,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 7300
    },
    {
      "epoch": 2.9168997203355973,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 7301
    },
    {
      "epoch": 2.917299240910907,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 7302
    },
    {
      "epoch": 2.9176987614862164,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 7303
    },
    {
      "epoch": 2.918098282061526,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 7304
    },
    {
      "epoch": 2.9184978026368356,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 7305
    },
    {
      "epoch": 2.9188973232121453,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 7306
    },
    {
      "epoch": 2.919296843787455,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 7307
    },
    {
      "epoch": 2.919696364362765,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 7308
    },
    {
      "epoch": 2.9200958849380743,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0687,
      "step": 7309
    },
    {
      "epoch": 2.920495405513384,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 7310
    },
    {
      "epoch": 2.9208949260886934,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 7311
    },
    {
      "epoch": 2.921294446664003,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 7312
    },
    {
      "epoch": 2.921693967239313,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 7313
    },
    {
      "epoch": 2.9220934878146223,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 7314
    },
    {
      "epoch": 2.922493008389932,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 7315
    },
    {
      "epoch": 2.922892528965242,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 7316
    },
    {
      "epoch": 2.923292049540551,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 7317
    },
    {
      "epoch": 2.923691570115861,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 7318
    },
    {
      "epoch": 2.924091090691171,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 7319
    },
    {
      "epoch": 2.92449061126648,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 7320
    },
    {
      "epoch": 2.92489013184179,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 7321
    },
    {
      "epoch": 2.9252896524170993,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 7322
    },
    {
      "epoch": 2.925689172992409,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 7323
    },
    {
      "epoch": 2.926088693567719,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 7324
    },
    {
      "epoch": 2.9264882141430286,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 7325
    },
    {
      "epoch": 2.926887734718338,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 7326
    },
    {
      "epoch": 2.9272872552936477,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 7327
    },
    {
      "epoch": 2.927686775868957,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 7328
    },
    {
      "epoch": 2.928086296444267,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 7329
    },
    {
      "epoch": 2.9284858170195767,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 7330
    },
    {
      "epoch": 2.928885337594886,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 7331
    },
    {
      "epoch": 2.929284858170196,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 7332
    },
    {
      "epoch": 2.929684378745505,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 7333
    },
    {
      "epoch": 2.930083899320815,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 7334
    },
    {
      "epoch": 2.9304834198961247,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 7335
    },
    {
      "epoch": 2.9308829404714345,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 7336
    },
    {
      "epoch": 2.931282461046744,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 7337
    },
    {
      "epoch": 2.9316819816220536,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 7338
    },
    {
      "epoch": 2.932081502197363,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 7339
    },
    {
      "epoch": 2.9324810227726728,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 7340
    },
    {
      "epoch": 2.9328805433479825,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 7341
    },
    {
      "epoch": 2.9332800639232923,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 7342
    },
    {
      "epoch": 2.9336795844986017,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 7343
    },
    {
      "epoch": 2.9340791050739115,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 7344
    },
    {
      "epoch": 2.934478625649221,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 7345
    },
    {
      "epoch": 2.9348781462245306,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 7346
    },
    {
      "epoch": 2.9352776667998404,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 7347
    },
    {
      "epoch": 2.9356771873751497,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 7348
    },
    {
      "epoch": 2.9360767079504595,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 7349
    },
    {
      "epoch": 2.936476228525769,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 7350
    },
    {
      "epoch": 2.9368757491010786,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 7351
    },
    {
      "epoch": 2.9372752696763884,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 7352
    },
    {
      "epoch": 2.937674790251698,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 7353
    },
    {
      "epoch": 2.9380743108270075,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 7354
    },
    {
      "epoch": 2.9384738314023173,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 7355
    },
    {
      "epoch": 2.9388733519776267,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 7356
    },
    {
      "epoch": 2.9392728725529365,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 7357
    },
    {
      "epoch": 2.9396723931282462,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 7358
    },
    {
      "epoch": 2.9400719137035556,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 7359
    },
    {
      "epoch": 2.9404714342788654,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 7360
    },
    {
      "epoch": 2.9408709548541747,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 7361
    },
    {
      "epoch": 2.9412704754294845,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0707,
      "step": 7362
    },
    {
      "epoch": 2.9416699960047943,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 7363
    },
    {
      "epoch": 2.942069516580104,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 7364
    },
    {
      "epoch": 2.9424690371554134,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 7365
    },
    {
      "epoch": 2.942868557730723,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 7366
    },
    {
      "epoch": 2.9432680783060325,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 7367
    },
    {
      "epoch": 2.9436675988813423,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 7368
    },
    {
      "epoch": 2.944067119456652,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 7369
    },
    {
      "epoch": 2.944466640031962,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 7370
    },
    {
      "epoch": 2.9448661606072712,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 7371
    },
    {
      "epoch": 2.945265681182581,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 7372
    },
    {
      "epoch": 2.9456652017578904,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 7373
    },
    {
      "epoch": 2.9460647223332,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 7374
    },
    {
      "epoch": 2.94646424290851,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 7375
    },
    {
      "epoch": 2.9468637634838193,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 7376
    },
    {
      "epoch": 2.947263284059129,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 7377
    },
    {
      "epoch": 2.9476628046344384,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 7378
    },
    {
      "epoch": 2.948062325209748,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 7379
    },
    {
      "epoch": 2.948461845785058,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 7380
    },
    {
      "epoch": 2.9488613663603678,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 7381
    },
    {
      "epoch": 2.949260886935677,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 7382
    },
    {
      "epoch": 2.949660407510987,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 7383
    },
    {
      "epoch": 2.9500599280862962,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 7384
    },
    {
      "epoch": 2.950459448661606,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 7385
    },
    {
      "epoch": 2.950858969236916,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 7386
    },
    {
      "epoch": 2.9512584898122256,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 7387
    },
    {
      "epoch": 2.951658010387535,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 7388
    },
    {
      "epoch": 2.9520575309628447,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 7389
    },
    {
      "epoch": 2.952457051538154,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 7390
    },
    {
      "epoch": 2.952856572113464,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 7391
    },
    {
      "epoch": 2.9532560926887736,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 7392
    },
    {
      "epoch": 2.953655613264083,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 7393
    },
    {
      "epoch": 2.9540551338393928,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 7394
    },
    {
      "epoch": 2.954454654414702,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 7395
    },
    {
      "epoch": 2.954854174990012,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 7396
    },
    {
      "epoch": 2.9552536955653217,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 7397
    },
    {
      "epoch": 2.9556532161406315,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 7398
    },
    {
      "epoch": 2.956052736715941,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 7399
    },
    {
      "epoch": 2.9564522572912506,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 7400
    },
    {
      "epoch": 2.95685177786656,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 7401
    },
    {
      "epoch": 2.9572512984418697,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 7402
    },
    {
      "epoch": 2.9576508190171795,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 7403
    },
    {
      "epoch": 2.958050339592489,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 7404
    },
    {
      "epoch": 2.9584498601677987,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 7405
    },
    {
      "epoch": 2.958849380743108,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 7406
    },
    {
      "epoch": 2.959248901318418,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 7407
    },
    {
      "epoch": 2.9596484218937276,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 7408
    },
    {
      "epoch": 2.9600479424690374,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 7409
    },
    {
      "epoch": 2.9604474630443467,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 7410
    },
    {
      "epoch": 2.9608469836196565,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 7411
    },
    {
      "epoch": 2.961246504194966,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 7412
    },
    {
      "epoch": 2.9616460247702756,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 7413
    },
    {
      "epoch": 2.9620455453455854,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 7414
    },
    {
      "epoch": 2.962445065920895,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 7415
    },
    {
      "epoch": 2.9628445864962045,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 7416
    },
    {
      "epoch": 2.9632441070715143,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0686,
      "step": 7417
    },
    {
      "epoch": 2.9636436276468237,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 7418
    },
    {
      "epoch": 2.9640431482221334,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 7419
    },
    {
      "epoch": 2.9644426687974432,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 7420
    },
    {
      "epoch": 2.9648421893727526,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 7421
    },
    {
      "epoch": 2.9652417099480624,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 7422
    },
    {
      "epoch": 2.9656412305233717,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 7423
    },
    {
      "epoch": 2.9660407510986815,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0663,
      "step": 7424
    },
    {
      "epoch": 2.9664402716739913,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 7425
    },
    {
      "epoch": 2.966839792249301,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 7426
    },
    {
      "epoch": 2.9672393128246104,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 7427
    },
    {
      "epoch": 2.96763883339992,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 7428
    },
    {
      "epoch": 2.9680383539752295,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 7429
    },
    {
      "epoch": 2.9684378745505393,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 7430
    },
    {
      "epoch": 2.968837395125849,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 7431
    },
    {
      "epoch": 2.969236915701159,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 7432
    },
    {
      "epoch": 2.9696364362764682,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 7433
    },
    {
      "epoch": 2.970035956851778,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 7434
    },
    {
      "epoch": 2.9704354774270874,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 7435
    },
    {
      "epoch": 2.970834998002397,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 7436
    },
    {
      "epoch": 2.971234518577707,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 7437
    },
    {
      "epoch": 2.9716340391530163,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 7438
    },
    {
      "epoch": 2.972033559728326,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 7439
    },
    {
      "epoch": 2.9724330803036354,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 7440
    },
    {
      "epoch": 2.972832600878945,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 7441
    },
    {
      "epoch": 2.973232121454255,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 7442
    },
    {
      "epoch": 2.9736316420295648,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 7443
    },
    {
      "epoch": 2.974031162604874,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 7444
    },
    {
      "epoch": 2.974430683180184,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 7445
    },
    {
      "epoch": 2.9748302037554932,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 7446
    },
    {
      "epoch": 2.975229724330803,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0657,
      "step": 7447
    },
    {
      "epoch": 2.975629244906113,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0674,
      "step": 7448
    },
    {
      "epoch": 2.976028765481422,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 7449
    },
    {
      "epoch": 2.976428286056732,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 7450
    },
    {
      "epoch": 2.9768278066320413,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 7451
    },
    {
      "epoch": 2.977227327207351,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 7452
    },
    {
      "epoch": 2.977626847782661,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 7453
    },
    {
      "epoch": 2.9780263683579706,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 7454
    },
    {
      "epoch": 2.97842588893328,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0747,
      "step": 7455
    },
    {
      "epoch": 2.9788254095085898,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 7456
    },
    {
      "epoch": 2.979224930083899,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 7457
    },
    {
      "epoch": 2.979624450659209,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 7458
    },
    {
      "epoch": 2.9800239712345187,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 7459
    },
    {
      "epoch": 2.9804234918098285,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 7460
    },
    {
      "epoch": 2.980823012385138,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 7461
    },
    {
      "epoch": 2.9812225329604476,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 7462
    },
    {
      "epoch": 2.981622053535757,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 7463
    },
    {
      "epoch": 2.9820215741110667,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 7464
    },
    {
      "epoch": 2.9824210946863765,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 7465
    },
    {
      "epoch": 2.982820615261686,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 7466
    },
    {
      "epoch": 2.9832201358369956,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 7467
    },
    {
      "epoch": 2.983619656412305,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 7468
    },
    {
      "epoch": 2.9840191769876148,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 7469
    },
    {
      "epoch": 2.9844186975629245,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 7470
    },
    {
      "epoch": 2.9848182181382343,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 7471
    },
    {
      "epoch": 2.9852177387135437,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 7472
    },
    {
      "epoch": 2.9856172592888535,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 7473
    },
    {
      "epoch": 2.986016779864163,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 7474
    },
    {
      "epoch": 2.9864163004394726,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 7475
    },
    {
      "epoch": 2.9868158210147824,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 7476
    },
    {
      "epoch": 2.987215341590092,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 7477
    },
    {
      "epoch": 2.9876148621654015,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 7478
    },
    {
      "epoch": 2.9880143827407113,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 7479
    },
    {
      "epoch": 2.9884139033160206,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 7480
    },
    {
      "epoch": 2.9888134238913304,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 7481
    },
    {
      "epoch": 2.98921294446664,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 7482
    },
    {
      "epoch": 2.9896124650419496,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 7483
    },
    {
      "epoch": 2.9900119856172593,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 7484
    },
    {
      "epoch": 2.9904115061925687,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 7485
    },
    {
      "epoch": 2.9908110267678785,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 7486
    },
    {
      "epoch": 2.9912105473431883,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 7487
    },
    {
      "epoch": 2.991610067918498,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 7488
    },
    {
      "epoch": 2.9920095884938074,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 7489
    },
    {
      "epoch": 2.992409109069117,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 7490
    },
    {
      "epoch": 2.9928086296444265,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 7491
    },
    {
      "epoch": 2.9932081502197363,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 7492
    },
    {
      "epoch": 2.993607670795046,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 7493
    },
    {
      "epoch": 2.9940071913703554,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 7494
    },
    {
      "epoch": 2.994406711945665,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 7495
    },
    {
      "epoch": 2.9948062325209746,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 7496
    },
    {
      "epoch": 2.9952057530962843,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 7497
    },
    {
      "epoch": 2.995605273671594,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 7498
    },
    {
      "epoch": 2.996004794246904,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 7499
    },
    {
      "epoch": 2.9964043148222133,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 7500
    },
    {
      "epoch": 2.996803835397523,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 7501
    },
    {
      "epoch": 2.9972033559728324,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 7502
    },
    {
      "epoch": 2.997602876548142,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 7503
    },
    {
      "epoch": 2.998002397123452,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 7504
    },
    {
      "epoch": 2.9984019176987617,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 7505
    },
    {
      "epoch": 2.998801438274071,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 7506
    },
    {
      "epoch": 2.999200958849381,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0713,
      "step": 7507
    },
    {
      "epoch": 2.99960047942469,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 7508
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 7509
    },
    {
      "epoch": 3.00039952057531,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 7510
    },
    {
      "epoch": 3.000799041150619,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 7511
    },
    {
      "epoch": 3.001198561725929,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 7512
    },
    {
      "epoch": 3.0015980823012387,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 7513
    },
    {
      "epoch": 3.001997602876548,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 7514
    },
    {
      "epoch": 3.002397123451858,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 7515
    },
    {
      "epoch": 3.0027966440271676,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 7516
    },
    {
      "epoch": 3.003196164602477,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 7517
    },
    {
      "epoch": 3.0035956851777867,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 7518
    },
    {
      "epoch": 3.003995205753096,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 7519
    },
    {
      "epoch": 3.004394726328406,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 7520
    },
    {
      "epoch": 3.0047942469037157,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 7521
    },
    {
      "epoch": 3.005193767479025,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 7522
    },
    {
      "epoch": 3.005593288054335,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 7523
    },
    {
      "epoch": 3.0059928086296446,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 7524
    },
    {
      "epoch": 3.006392329204954,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 7525
    },
    {
      "epoch": 3.0067918497802637,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 7526
    },
    {
      "epoch": 3.0071913703555735,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 7527
    },
    {
      "epoch": 3.007590890930883,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 7528
    },
    {
      "epoch": 3.0079904115061926,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 7529
    },
    {
      "epoch": 3.0083899320815024,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 7530
    },
    {
      "epoch": 3.0087894526568117,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 7531
    },
    {
      "epoch": 3.0091889732321215,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 7532
    },
    {
      "epoch": 3.009588493807431,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 7533
    },
    {
      "epoch": 3.0099880143827407,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 7534
    },
    {
      "epoch": 3.0103875349580504,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 7535
    },
    {
      "epoch": 3.01078705553336,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 7536
    },
    {
      "epoch": 3.0111865761086696,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 7537
    },
    {
      "epoch": 3.0115860966839794,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 7538
    },
    {
      "epoch": 3.0119856172592887,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 7539
    },
    {
      "epoch": 3.0123851378345985,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 7540
    },
    {
      "epoch": 3.0127846584099083,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 7541
    },
    {
      "epoch": 3.0131841789852176,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 7542
    },
    {
      "epoch": 3.0135836995605274,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 7543
    },
    {
      "epoch": 3.013983220135837,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 7544
    },
    {
      "epoch": 3.0143827407111465,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 7545
    },
    {
      "epoch": 3.0147822612864563,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 7546
    },
    {
      "epoch": 3.0151817818617657,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 7547
    },
    {
      "epoch": 3.0155813024370755,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 7548
    },
    {
      "epoch": 3.0159808230123852,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 7549
    },
    {
      "epoch": 3.0163803435876946,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 7550
    },
    {
      "epoch": 3.0167798641630044,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 7551
    },
    {
      "epoch": 3.017179384738314,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 7552
    },
    {
      "epoch": 3.0175789053136235,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 7553
    },
    {
      "epoch": 3.0179784258889333,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 7554
    },
    {
      "epoch": 3.018377946464243,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 7555
    },
    {
      "epoch": 3.0187774670395524,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 7556
    },
    {
      "epoch": 3.019176987614862,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 7557
    },
    {
      "epoch": 3.019576508190172,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 7558
    },
    {
      "epoch": 3.0199760287654813,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 7559
    },
    {
      "epoch": 3.020375549340791,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 7560
    },
    {
      "epoch": 3.020775069916101,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 7561
    },
    {
      "epoch": 3.0211745904914102,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 7562
    },
    {
      "epoch": 3.02157411106672,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 7563
    },
    {
      "epoch": 3.0219736316420294,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 7564
    },
    {
      "epoch": 3.022373152217339,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 7565
    },
    {
      "epoch": 3.022772672792649,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 7566
    },
    {
      "epoch": 3.0231721933679583,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 7567
    },
    {
      "epoch": 3.023571713943268,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 7568
    },
    {
      "epoch": 3.023971234518578,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 7569
    },
    {
      "epoch": 3.024370755093887,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 7570
    },
    {
      "epoch": 3.024770275669197,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 7571
    },
    {
      "epoch": 3.0251697962445068,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 7572
    },
    {
      "epoch": 3.025569316819816,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 7573
    },
    {
      "epoch": 3.025968837395126,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 7574
    },
    {
      "epoch": 3.0263683579704357,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 7575
    },
    {
      "epoch": 3.026767878545745,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 7576
    },
    {
      "epoch": 3.027167399121055,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 7577
    },
    {
      "epoch": 3.027566919696364,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 7578
    },
    {
      "epoch": 3.027966440271674,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 7579
    },
    {
      "epoch": 3.0283659608469837,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 7580
    },
    {
      "epoch": 3.028765481422293,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 7581
    },
    {
      "epoch": 3.029165001997603,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 7582
    },
    {
      "epoch": 3.0295645225729126,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 7583
    },
    {
      "epoch": 3.029964043148222,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 7584
    },
    {
      "epoch": 3.0303635637235318,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 7585
    },
    {
      "epoch": 3.0307630842988416,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 7586
    },
    {
      "epoch": 3.031162604874151,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 7587
    },
    {
      "epoch": 3.0315621254494607,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 7588
    },
    {
      "epoch": 3.0319616460247705,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 7589
    },
    {
      "epoch": 3.03236116660008,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 7590
    },
    {
      "epoch": 3.0327606871753896,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0697,
      "step": 7591
    },
    {
      "epoch": 3.033160207750699,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 7592
    },
    {
      "epoch": 3.0335597283260087,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 7593
    },
    {
      "epoch": 3.0339592489013185,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 7594
    },
    {
      "epoch": 3.034358769476628,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 7595
    },
    {
      "epoch": 3.0347582900519376,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 7596
    },
    {
      "epoch": 3.0351578106272474,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 7597
    },
    {
      "epoch": 3.0355573312025568,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 7598
    },
    {
      "epoch": 3.0359568517778666,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 7599
    },
    {
      "epoch": 3.0363563723531763,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 7600
    },
    {
      "epoch": 3.0367558929284857,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 7601
    },
    {
      "epoch": 3.0371554135037955,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 7602
    },
    {
      "epoch": 3.0375549340791053,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 7603
    },
    {
      "epoch": 3.0379544546544146,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 7604
    },
    {
      "epoch": 3.0383539752297244,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 7605
    },
    {
      "epoch": 3.038753495805034,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 7606
    },
    {
      "epoch": 3.0391530163803435,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 7607
    },
    {
      "epoch": 3.0395525369556533,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 7608
    },
    {
      "epoch": 3.0399520575309626,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 7609
    },
    {
      "epoch": 3.0403515781062724,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 7610
    },
    {
      "epoch": 3.040751098681582,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 7611
    },
    {
      "epoch": 3.0411506192568916,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 7612
    },
    {
      "epoch": 3.0415501398322013,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 7613
    },
    {
      "epoch": 3.041949660407511,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 7614
    },
    {
      "epoch": 3.0423491809828205,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 7615
    },
    {
      "epoch": 3.0427487015581303,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0598,
      "step": 7616
    },
    {
      "epoch": 3.04314822213344,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 7617
    },
    {
      "epoch": 3.0435477427087494,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 7618
    },
    {
      "epoch": 3.043947263284059,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 7619
    },
    {
      "epoch": 3.044346783859369,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 7620
    },
    {
      "epoch": 3.0447463044346783,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 7621
    },
    {
      "epoch": 3.045145825009988,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 7622
    },
    {
      "epoch": 3.0455453455852974,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 7623
    },
    {
      "epoch": 3.0459448661606072,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 7624
    },
    {
      "epoch": 3.046344386735917,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 7625
    },
    {
      "epoch": 3.0467439073112264,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 7626
    },
    {
      "epoch": 3.047143427886536,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 7627
    },
    {
      "epoch": 3.047542948461846,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 7628
    },
    {
      "epoch": 3.0479424690371553,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 7629
    },
    {
      "epoch": 3.048341989612465,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 7630
    },
    {
      "epoch": 3.048741510187775,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 7631
    },
    {
      "epoch": 3.049141030763084,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 7632
    },
    {
      "epoch": 3.049540551338394,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 7633
    },
    {
      "epoch": 3.0499400719137038,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 7634
    },
    {
      "epoch": 3.050339592489013,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 7635
    },
    {
      "epoch": 3.050739113064323,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 7636
    },
    {
      "epoch": 3.0511386336396322,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 7637
    },
    {
      "epoch": 3.051538154214942,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 7638
    },
    {
      "epoch": 3.051937674790252,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 7639
    },
    {
      "epoch": 3.052337195365561,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 7640
    },
    {
      "epoch": 3.052736715940871,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 7641
    },
    {
      "epoch": 3.0531362365161807,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 7642
    },
    {
      "epoch": 3.05353575709149,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 7643
    },
    {
      "epoch": 3.0539352776668,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 7644
    },
    {
      "epoch": 3.0543347982421096,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 7645
    },
    {
      "epoch": 3.054734318817419,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 7646
    },
    {
      "epoch": 3.0551338393927288,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 7647
    },
    {
      "epoch": 3.0555333599680385,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 7648
    },
    {
      "epoch": 3.055932880543348,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 7649
    },
    {
      "epoch": 3.0563324011186577,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 7650
    },
    {
      "epoch": 3.0567319216939675,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 7651
    },
    {
      "epoch": 3.057131442269277,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 7652
    },
    {
      "epoch": 3.0575309628445866,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 7653
    },
    {
      "epoch": 3.057930483419896,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 7654
    },
    {
      "epoch": 3.0583300039952057,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 7655
    },
    {
      "epoch": 3.0587295245705155,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 7656
    },
    {
      "epoch": 3.059129045145825,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 7657
    },
    {
      "epoch": 3.0595285657211346,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 7658
    },
    {
      "epoch": 3.0599280862964444,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 7659
    },
    {
      "epoch": 3.0603276068717538,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 7660
    },
    {
      "epoch": 3.0607271274470635,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 7661
    },
    {
      "epoch": 3.0611266480223733,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 7662
    },
    {
      "epoch": 3.0615261685976827,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 7663
    },
    {
      "epoch": 3.0619256891729925,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 7664
    },
    {
      "epoch": 3.0623252097483022,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 7665
    },
    {
      "epoch": 3.0627247303236116,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 7666
    },
    {
      "epoch": 3.0631242508989214,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 7667
    },
    {
      "epoch": 3.0635237714742307,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 7668
    },
    {
      "epoch": 3.0639232920495405,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 7669
    },
    {
      "epoch": 3.0643228126248503,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 7670
    },
    {
      "epoch": 3.0647223332001596,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 7671
    },
    {
      "epoch": 3.0651218537754694,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 7672
    },
    {
      "epoch": 3.065521374350779,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 7673
    },
    {
      "epoch": 3.0659208949260885,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 7674
    },
    {
      "epoch": 3.0663204155013983,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 7675
    },
    {
      "epoch": 3.066719936076708,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 7676
    },
    {
      "epoch": 3.0671194566520175,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 7677
    },
    {
      "epoch": 3.0675189772273272,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 7678
    },
    {
      "epoch": 3.067918497802637,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 7679
    },
    {
      "epoch": 3.0683180183779464,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 7680
    },
    {
      "epoch": 3.068717538953256,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 7681
    },
    {
      "epoch": 3.0691170595285655,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 7682
    },
    {
      "epoch": 3.0695165801038753,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 7683
    },
    {
      "epoch": 3.069916100679185,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 7684
    },
    {
      "epoch": 3.0703156212544944,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 7685
    },
    {
      "epoch": 3.070715141829804,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 7686
    },
    {
      "epoch": 3.071114662405114,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 7687
    },
    {
      "epoch": 3.0715141829804233,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 7688
    },
    {
      "epoch": 3.071913703555733,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 7689
    },
    {
      "epoch": 3.072313224131043,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 7690
    },
    {
      "epoch": 3.0727127447063523,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 7691
    },
    {
      "epoch": 3.073112265281662,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 7692
    },
    {
      "epoch": 3.073511785856972,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 7693
    },
    {
      "epoch": 3.073911306432281,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 7694
    },
    {
      "epoch": 3.074310827007591,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 7695
    },
    {
      "epoch": 3.0747103475829007,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0662,
      "step": 7696
    },
    {
      "epoch": 3.07510986815821,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 7697
    },
    {
      "epoch": 3.07550938873352,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 7698
    },
    {
      "epoch": 3.075908909308829,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 7699
    },
    {
      "epoch": 3.076308429884139,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 7700
    },
    {
      "epoch": 3.076707950459449,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 7701
    },
    {
      "epoch": 3.077107471034758,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 7702
    },
    {
      "epoch": 3.077506991610068,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 7703
    },
    {
      "epoch": 3.0779065121853777,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 7704
    },
    {
      "epoch": 3.078306032760687,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 7705
    },
    {
      "epoch": 3.078705553335997,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 7706
    },
    {
      "epoch": 3.0791050739113066,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 7707
    },
    {
      "epoch": 3.079504594486616,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 7708
    },
    {
      "epoch": 3.0799041150619257,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 7709
    },
    {
      "epoch": 3.0803036356372355,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 7710
    },
    {
      "epoch": 3.080703156212545,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 7711
    },
    {
      "epoch": 3.0811026767878547,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 7712
    },
    {
      "epoch": 3.081502197363164,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 7713
    },
    {
      "epoch": 3.081901717938474,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 7714
    },
    {
      "epoch": 3.0823012385137836,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 7715
    },
    {
      "epoch": 3.082700759089093,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 7716
    },
    {
      "epoch": 3.0831002796644027,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 7717
    },
    {
      "epoch": 3.0834998002397125,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 7718
    },
    {
      "epoch": 3.083899320815022,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 7719
    },
    {
      "epoch": 3.0842988413903316,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 7720
    },
    {
      "epoch": 3.0846983619656414,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 7721
    },
    {
      "epoch": 3.0850978825409507,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 7722
    },
    {
      "epoch": 3.0854974031162605,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 7723
    },
    {
      "epoch": 3.0858969236915703,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0683,
      "step": 7724
    },
    {
      "epoch": 3.0862964442668797,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 7725
    },
    {
      "epoch": 3.0866959648421894,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 7726
    },
    {
      "epoch": 3.087095485417499,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 7727
    },
    {
      "epoch": 3.0874950059928086,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 7728
    },
    {
      "epoch": 3.0878945265681184,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 7729
    },
    {
      "epoch": 3.0882940471434277,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 7730
    },
    {
      "epoch": 3.0886935677187375,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 7731
    },
    {
      "epoch": 3.0890930882940473,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 7732
    },
    {
      "epoch": 3.0894926088693566,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 7733
    },
    {
      "epoch": 3.0898921294446664,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 7734
    },
    {
      "epoch": 3.090291650019976,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 7735
    },
    {
      "epoch": 3.0906911705952855,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 7736
    },
    {
      "epoch": 3.0910906911705953,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 7737
    },
    {
      "epoch": 3.091490211745905,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 7738
    },
    {
      "epoch": 3.0918897323212144,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 7739
    },
    {
      "epoch": 3.0922892528965242,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 7740
    },
    {
      "epoch": 3.092688773471834,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 7741
    },
    {
      "epoch": 3.0930882940471434,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 7742
    },
    {
      "epoch": 3.093487814622453,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 7743
    },
    {
      "epoch": 3.0938873351977625,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0659,
      "step": 7744
    },
    {
      "epoch": 3.0942868557730723,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 7745
    },
    {
      "epoch": 3.094686376348382,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 7746
    },
    {
      "epoch": 3.0950858969236914,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 7747
    },
    {
      "epoch": 3.095485417499001,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 7748
    },
    {
      "epoch": 3.095884938074311,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 7749
    },
    {
      "epoch": 3.0962844586496203,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 7750
    },
    {
      "epoch": 3.09668397922493,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 7751
    },
    {
      "epoch": 3.09708349980024,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 7752
    },
    {
      "epoch": 3.0974830203755492,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0723,
      "step": 7753
    },
    {
      "epoch": 3.097882540950859,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 7754
    },
    {
      "epoch": 3.098282061526169,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 7755
    },
    {
      "epoch": 3.098681582101478,
      "grad_norm": 0.21875,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 7756
    },
    {
      "epoch": 3.099081102676788,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 7757
    },
    {
      "epoch": 3.0994806232520973,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 7758
    },
    {
      "epoch": 3.099880143827407,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 7759
    },
    {
      "epoch": 3.100279664402717,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 7760
    },
    {
      "epoch": 3.100679184978026,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 7761
    },
    {
      "epoch": 3.101078705553336,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 7762
    },
    {
      "epoch": 3.1014782261286458,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 7763
    },
    {
      "epoch": 3.101877746703955,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 7764
    },
    {
      "epoch": 3.102277267279265,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 7765
    },
    {
      "epoch": 3.1026767878545747,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 7766
    },
    {
      "epoch": 3.103076308429884,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 7767
    },
    {
      "epoch": 3.103475829005194,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 7768
    },
    {
      "epoch": 3.1038753495805036,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0598,
      "step": 7769
    },
    {
      "epoch": 3.104274870155813,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 7770
    },
    {
      "epoch": 3.1046743907311227,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 7771
    },
    {
      "epoch": 3.105073911306432,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 7772
    },
    {
      "epoch": 3.105473431881742,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 7773
    },
    {
      "epoch": 3.1058729524570516,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 7774
    },
    {
      "epoch": 3.106272473032361,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 7775
    },
    {
      "epoch": 3.1066719936076708,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 7776
    },
    {
      "epoch": 3.1070715141829806,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 7777
    },
    {
      "epoch": 3.10747103475829,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 7778
    },
    {
      "epoch": 3.1078705553335997,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 7779
    },
    {
      "epoch": 3.1082700759089095,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 7780
    },
    {
      "epoch": 3.108669596484219,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 7781
    },
    {
      "epoch": 3.1090691170595286,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0702,
      "step": 7782
    },
    {
      "epoch": 3.1094686376348384,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0721,
      "step": 7783
    },
    {
      "epoch": 3.1098681582101477,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 7784
    },
    {
      "epoch": 3.1102676787854575,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 7785
    },
    {
      "epoch": 3.1106671993607673,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 7786
    },
    {
      "epoch": 3.1110667199360766,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 7787
    },
    {
      "epoch": 3.1114662405113864,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 7788
    },
    {
      "epoch": 3.1118657610866958,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 7789
    },
    {
      "epoch": 3.1122652816620056,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 7790
    },
    {
      "epoch": 3.1126648022373153,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 7791
    },
    {
      "epoch": 3.1130643228126247,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 7792
    },
    {
      "epoch": 3.1134638433879345,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 7793
    },
    {
      "epoch": 3.1138633639632443,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 7794
    },
    {
      "epoch": 3.1142628845385536,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 7795
    },
    {
      "epoch": 3.1146624051138634,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 7796
    },
    {
      "epoch": 3.115061925689173,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 7797
    },
    {
      "epoch": 3.1154614462644825,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 7798
    },
    {
      "epoch": 3.1158609668397923,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 7799
    },
    {
      "epoch": 3.116260487415102,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 7800
    },
    {
      "epoch": 3.1166600079904114,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 7801
    },
    {
      "epoch": 3.117059528565721,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 7802
    },
    {
      "epoch": 3.1174590491410306,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 7803
    },
    {
      "epoch": 3.1178585697163403,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 7804
    },
    {
      "epoch": 3.11825809029165,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 7805
    },
    {
      "epoch": 3.1186576108669595,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 7806
    },
    {
      "epoch": 3.1190571314422693,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 7807
    },
    {
      "epoch": 3.119456652017579,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 7808
    },
    {
      "epoch": 3.1198561725928884,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 7809
    },
    {
      "epoch": 3.120255693168198,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 7810
    },
    {
      "epoch": 3.120655213743508,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 7811
    },
    {
      "epoch": 3.1210547343188173,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 7812
    },
    {
      "epoch": 3.121454254894127,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 7813
    },
    {
      "epoch": 3.121853775469437,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 7814
    },
    {
      "epoch": 3.122253296044746,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 7815
    },
    {
      "epoch": 3.122652816620056,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 7816
    },
    {
      "epoch": 3.1230523371953653,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 7817
    },
    {
      "epoch": 3.123451857770675,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 7818
    },
    {
      "epoch": 3.123851378345985,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 7819
    },
    {
      "epoch": 3.1242508989212943,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 7820
    },
    {
      "epoch": 3.124650419496604,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 7821
    },
    {
      "epoch": 3.125049940071914,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 7822
    },
    {
      "epoch": 3.125449460647223,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 7823
    },
    {
      "epoch": 3.125848981222533,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 7824
    },
    {
      "epoch": 3.1262485017978427,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 7825
    },
    {
      "epoch": 3.126648022373152,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 7826
    },
    {
      "epoch": 3.127047542948462,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 7827
    },
    {
      "epoch": 3.1274470635237717,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 7828
    },
    {
      "epoch": 3.127846584099081,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 7829
    },
    {
      "epoch": 3.128246104674391,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 7830
    },
    {
      "epoch": 3.1286456252497006,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 7831
    },
    {
      "epoch": 3.12904514582501,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 7832
    },
    {
      "epoch": 3.1294446664003197,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 7833
    },
    {
      "epoch": 3.129844186975629,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 7834
    },
    {
      "epoch": 3.130243707550939,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 7835
    },
    {
      "epoch": 3.1306432281262486,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 7836
    },
    {
      "epoch": 3.131042748701558,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 7837
    },
    {
      "epoch": 3.1314422692768678,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 7838
    },
    {
      "epoch": 3.1318417898521775,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 7839
    },
    {
      "epoch": 3.132241310427487,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 7840
    },
    {
      "epoch": 3.1326408310027967,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 7841
    },
    {
      "epoch": 3.1330403515781065,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 7842
    },
    {
      "epoch": 3.133439872153416,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 7843
    },
    {
      "epoch": 3.1338393927287256,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 7844
    },
    {
      "epoch": 3.134238913304035,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 7845
    },
    {
      "epoch": 3.1346384338793447,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 7846
    },
    {
      "epoch": 3.1350379544546545,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 7847
    },
    {
      "epoch": 3.135437475029964,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 7848
    },
    {
      "epoch": 3.1358369956052736,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 7849
    },
    {
      "epoch": 3.1362365161805834,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 7850
    },
    {
      "epoch": 3.1366360367558928,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 7851
    },
    {
      "epoch": 3.1370355573312025,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 7852
    },
    {
      "epoch": 3.1374350779065123,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 7853
    },
    {
      "epoch": 3.1378345984818217,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 7854
    },
    {
      "epoch": 3.1382341190571315,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 7855
    },
    {
      "epoch": 3.1386336396324412,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 7856
    },
    {
      "epoch": 3.1390331602077506,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 7857
    },
    {
      "epoch": 3.1394326807830604,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 7858
    },
    {
      "epoch": 3.13983220135837,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 7859
    },
    {
      "epoch": 3.1402317219336795,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 7860
    },
    {
      "epoch": 3.1406312425089893,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 7861
    },
    {
      "epoch": 3.1410307630842986,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 7862
    },
    {
      "epoch": 3.1414302836596084,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 7863
    },
    {
      "epoch": 3.141829804234918,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 7864
    },
    {
      "epoch": 3.1422293248102275,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 7865
    },
    {
      "epoch": 3.1426288453855373,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 7866
    },
    {
      "epoch": 3.143028365960847,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 7867
    },
    {
      "epoch": 3.1434278865361565,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 7868
    },
    {
      "epoch": 3.1438274071114662,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 7869
    },
    {
      "epoch": 3.144226927686776,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 7870
    },
    {
      "epoch": 3.1446264482620854,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 7871
    },
    {
      "epoch": 3.145025968837395,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 7872
    },
    {
      "epoch": 3.145425489412705,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 7873
    },
    {
      "epoch": 3.1458250099880143,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 7874
    },
    {
      "epoch": 3.146224530563324,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 7875
    },
    {
      "epoch": 3.146624051138634,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 7876
    },
    {
      "epoch": 3.147023571713943,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 7877
    },
    {
      "epoch": 3.147423092289253,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 7878
    },
    {
      "epoch": 3.1478226128645623,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 7879
    },
    {
      "epoch": 3.148222133439872,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 7880
    },
    {
      "epoch": 3.148621654015182,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 7881
    },
    {
      "epoch": 3.1490211745904912,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 7882
    },
    {
      "epoch": 3.149420695165801,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 7883
    },
    {
      "epoch": 3.149820215741111,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 7884
    },
    {
      "epoch": 3.15021973631642,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 7885
    },
    {
      "epoch": 3.15061925689173,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 7886
    },
    {
      "epoch": 3.1510187774670397,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 7887
    },
    {
      "epoch": 3.151418298042349,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 7888
    },
    {
      "epoch": 3.151817818617659,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 7889
    },
    {
      "epoch": 3.152217339192968,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 7890
    },
    {
      "epoch": 3.152616859768278,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 7891
    },
    {
      "epoch": 3.1530163803435878,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 7892
    },
    {
      "epoch": 3.153415900918897,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 7893
    },
    {
      "epoch": 3.153815421494207,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 7894
    },
    {
      "epoch": 3.1542149420695167,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 7895
    },
    {
      "epoch": 3.154614462644826,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 7896
    },
    {
      "epoch": 3.155013983220136,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 7897
    },
    {
      "epoch": 3.1554135037954456,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 7898
    },
    {
      "epoch": 3.155813024370755,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 7899
    },
    {
      "epoch": 3.1562125449460647,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 7900
    },
    {
      "epoch": 3.1566120655213745,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 7901
    },
    {
      "epoch": 3.157011586096684,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 7902
    },
    {
      "epoch": 3.1574111066719937,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 7903
    },
    {
      "epoch": 3.1578106272473034,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 7904
    },
    {
      "epoch": 3.158210147822613,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 7905
    },
    {
      "epoch": 3.1586096683979226,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 7906
    },
    {
      "epoch": 3.159009188973232,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 7907
    },
    {
      "epoch": 3.1594087095485417,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 7908
    },
    {
      "epoch": 3.1598082301238515,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0705,
      "step": 7909
    },
    {
      "epoch": 3.160207750699161,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 7910
    },
    {
      "epoch": 3.1606072712744706,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 7911
    },
    {
      "epoch": 3.1610067918497804,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 7912
    },
    {
      "epoch": 3.1614063124250897,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 7913
    },
    {
      "epoch": 3.1618058330003995,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 7914
    },
    {
      "epoch": 3.1622053535757093,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 7915
    },
    {
      "epoch": 3.1626048741510187,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 7916
    },
    {
      "epoch": 3.1630043947263284,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 7917
    },
    {
      "epoch": 3.1634039153016382,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 7918
    },
    {
      "epoch": 3.1638034358769476,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 7919
    },
    {
      "epoch": 3.1642029564522574,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 7920
    },
    {
      "epoch": 3.164602477027567,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0694,
      "step": 7921
    },
    {
      "epoch": 3.1650019976028765,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 7922
    },
    {
      "epoch": 3.1654015181781863,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 7923
    },
    {
      "epoch": 3.1658010387534956,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 7924
    },
    {
      "epoch": 3.1662005593288054,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 7925
    },
    {
      "epoch": 3.166600079904115,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 7926
    },
    {
      "epoch": 3.1669996004794245,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 7927
    },
    {
      "epoch": 3.1673991210547343,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 7928
    },
    {
      "epoch": 3.167798641630044,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 7929
    },
    {
      "epoch": 3.1681981622053534,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 7930
    },
    {
      "epoch": 3.1685976827806632,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0651,
      "step": 7931
    },
    {
      "epoch": 3.168997203355973,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 7932
    },
    {
      "epoch": 3.1693967239312824,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 7933
    },
    {
      "epoch": 3.169796244506592,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 7934
    },
    {
      "epoch": 3.1701957650819015,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 7935
    },
    {
      "epoch": 3.1705952856572113,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 7936
    },
    {
      "epoch": 3.170994806232521,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 7937
    },
    {
      "epoch": 3.1713943268078304,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 7938
    },
    {
      "epoch": 3.17179384738314,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 7939
    },
    {
      "epoch": 3.17219336795845,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 7940
    },
    {
      "epoch": 3.1725928885337593,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 7941
    },
    {
      "epoch": 3.172992409109069,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 7942
    },
    {
      "epoch": 3.173391929684379,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 7943
    },
    {
      "epoch": 3.1737914502596882,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 7944
    },
    {
      "epoch": 3.174190970834998,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 7945
    },
    {
      "epoch": 3.174590491410308,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 7946
    },
    {
      "epoch": 3.174990011985617,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 7947
    },
    {
      "epoch": 3.175389532560927,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 7948
    },
    {
      "epoch": 3.1757890531362367,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 7949
    },
    {
      "epoch": 3.176188573711546,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 7950
    },
    {
      "epoch": 3.176588094286856,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 7951
    },
    {
      "epoch": 3.176987614862165,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 7952
    },
    {
      "epoch": 3.177387135437475,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 7953
    },
    {
      "epoch": 3.1777866560127848,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 7954
    },
    {
      "epoch": 3.178186176588094,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 7955
    },
    {
      "epoch": 3.178585697163404,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 7956
    },
    {
      "epoch": 3.1789852177387137,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 7957
    },
    {
      "epoch": 3.179384738314023,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 7958
    },
    {
      "epoch": 3.179784258889333,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 7959
    },
    {
      "epoch": 3.1801837794646426,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0677,
      "step": 7960
    },
    {
      "epoch": 3.180583300039952,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 7961
    },
    {
      "epoch": 3.1809828206152617,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 7962
    },
    {
      "epoch": 3.1813823411905715,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 7963
    },
    {
      "epoch": 3.181781861765881,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 7964
    },
    {
      "epoch": 3.1821813823411906,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 7965
    },
    {
      "epoch": 3.1825809029165004,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 7966
    },
    {
      "epoch": 3.1829804234918098,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 7967
    },
    {
      "epoch": 3.1833799440671195,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 7968
    },
    {
      "epoch": 3.183779464642429,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 7969
    },
    {
      "epoch": 3.1841789852177387,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 7970
    },
    {
      "epoch": 3.1845785057930485,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 7971
    },
    {
      "epoch": 3.184978026368358,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 7972
    },
    {
      "epoch": 3.1853775469436676,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 7973
    },
    {
      "epoch": 3.1857770675189774,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 7974
    },
    {
      "epoch": 3.1861765880942867,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 7975
    },
    {
      "epoch": 3.1865761086695965,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 7976
    },
    {
      "epoch": 3.1869756292449063,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 7977
    },
    {
      "epoch": 3.1873751498202156,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 7978
    },
    {
      "epoch": 3.1877746703955254,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 7979
    },
    {
      "epoch": 3.1881741909708348,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 7980
    },
    {
      "epoch": 3.1885737115461446,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 7981
    },
    {
      "epoch": 3.1889732321214543,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 7982
    },
    {
      "epoch": 3.1893727526967637,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 7983
    },
    {
      "epoch": 3.1897722732720735,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 7984
    },
    {
      "epoch": 3.1901717938473833,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 7985
    },
    {
      "epoch": 3.1905713144226926,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 7986
    },
    {
      "epoch": 3.1909708349980024,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 7987
    },
    {
      "epoch": 3.191370355573312,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 7988
    },
    {
      "epoch": 3.1917698761486215,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 7989
    },
    {
      "epoch": 3.1921693967239313,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 7990
    },
    {
      "epoch": 3.192568917299241,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 7991
    },
    {
      "epoch": 3.1929684378745504,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 7992
    },
    {
      "epoch": 3.19336795844986,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 7993
    },
    {
      "epoch": 3.19376747902517,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 7994
    },
    {
      "epoch": 3.1941669996004793,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 7995
    },
    {
      "epoch": 3.194566520175789,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 7996
    },
    {
      "epoch": 3.1949660407510985,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 7997
    },
    {
      "epoch": 3.1953655613264083,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 7998
    },
    {
      "epoch": 3.195765081901718,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 7999
    },
    {
      "epoch": 3.1961646024770274,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 8000
    },
    {
      "epoch": 3.196564123052337,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 8001
    },
    {
      "epoch": 3.196963643627647,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 8002
    },
    {
      "epoch": 3.1973631642029563,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 8003
    },
    {
      "epoch": 3.197762684778266,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 8004
    },
    {
      "epoch": 3.198162205353576,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 8005
    },
    {
      "epoch": 3.198561725928885,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 8006
    },
    {
      "epoch": 3.198961246504195,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 8007
    },
    {
      "epoch": 3.199360767079505,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 8008
    },
    {
      "epoch": 3.199760287654814,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 8009
    },
    {
      "epoch": 3.200159808230124,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 8010
    },
    {
      "epoch": 3.2005593288054337,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 8011
    },
    {
      "epoch": 3.200958849380743,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 8012
    },
    {
      "epoch": 3.201358369956053,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 8013
    },
    {
      "epoch": 3.201757890531362,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 8014
    },
    {
      "epoch": 3.202157411106672,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 8015
    },
    {
      "epoch": 3.2025569316819817,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 8016
    },
    {
      "epoch": 3.202956452257291,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 8017
    },
    {
      "epoch": 3.203355972832601,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 8018
    },
    {
      "epoch": 3.2037554934079107,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 8019
    },
    {
      "epoch": 3.20415501398322,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 8020
    },
    {
      "epoch": 3.20455453455853,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 8021
    },
    {
      "epoch": 3.2049540551338396,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0676,
      "step": 8022
    },
    {
      "epoch": 3.205353575709149,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 8023
    },
    {
      "epoch": 3.2057530962844587,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 8024
    },
    {
      "epoch": 3.206152616859768,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 8025
    },
    {
      "epoch": 3.206552137435078,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 8026
    },
    {
      "epoch": 3.2069516580103876,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 8027
    },
    {
      "epoch": 3.207351178585697,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 8028
    },
    {
      "epoch": 3.2077506991610067,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 8029
    },
    {
      "epoch": 3.2081502197363165,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 8030
    },
    {
      "epoch": 3.208549740311626,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 8031
    },
    {
      "epoch": 3.2089492608869357,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 8032
    },
    {
      "epoch": 3.2093487814622454,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 8033
    },
    {
      "epoch": 3.209748302037555,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 8034
    },
    {
      "epoch": 3.2101478226128646,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 8035
    },
    {
      "epoch": 3.2105473431881744,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 8036
    },
    {
      "epoch": 3.2109468637634837,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 8037
    },
    {
      "epoch": 3.2113463843387935,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 8038
    },
    {
      "epoch": 3.2117459049141033,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 8039
    },
    {
      "epoch": 3.2121454254894126,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 8040
    },
    {
      "epoch": 3.2125449460647224,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 8041
    },
    {
      "epoch": 3.2129444666400317,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 8042
    },
    {
      "epoch": 3.2133439872153415,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 8043
    },
    {
      "epoch": 3.2137435077906513,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 8044
    },
    {
      "epoch": 3.2141430283659607,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 8045
    },
    {
      "epoch": 3.2145425489412705,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 8046
    },
    {
      "epoch": 3.2149420695165802,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 8047
    },
    {
      "epoch": 3.2153415900918896,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 8048
    },
    {
      "epoch": 3.2157411106671994,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 8049
    },
    {
      "epoch": 3.216140631242509,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 8050
    },
    {
      "epoch": 3.2165401518178185,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 8051
    },
    {
      "epoch": 3.2169396723931283,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 8052
    },
    {
      "epoch": 3.217339192968438,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 8053
    },
    {
      "epoch": 3.2177387135437474,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 8054
    },
    {
      "epoch": 3.218138234119057,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 8055
    },
    {
      "epoch": 3.218537754694367,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 8056
    },
    {
      "epoch": 3.2189372752696763,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 8057
    },
    {
      "epoch": 3.219336795844986,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 8058
    },
    {
      "epoch": 3.2197363164202955,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 8059
    },
    {
      "epoch": 3.2201358369956052,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 8060
    },
    {
      "epoch": 3.220535357570915,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 8061
    },
    {
      "epoch": 3.2209348781462244,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 8062
    },
    {
      "epoch": 3.221334398721534,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 8063
    },
    {
      "epoch": 3.221733919296844,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 8064
    },
    {
      "epoch": 3.2221334398721533,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 8065
    },
    {
      "epoch": 3.222532960447463,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 8066
    },
    {
      "epoch": 3.222932481022773,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 8067
    },
    {
      "epoch": 3.223332001598082,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.071,
      "step": 8068
    },
    {
      "epoch": 3.223731522173392,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 8069
    },
    {
      "epoch": 3.2241310427487013,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 8070
    },
    {
      "epoch": 3.224530563324011,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 8071
    },
    {
      "epoch": 3.224930083899321,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 8072
    },
    {
      "epoch": 3.2253296044746302,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 8073
    },
    {
      "epoch": 3.22572912504994,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 8074
    },
    {
      "epoch": 3.22612864562525,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 8075
    },
    {
      "epoch": 3.226528166200559,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 8076
    },
    {
      "epoch": 3.226927686775869,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 8077
    },
    {
      "epoch": 3.2273272073511787,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0651,
      "step": 8078
    },
    {
      "epoch": 3.227726727926488,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 8079
    },
    {
      "epoch": 3.228126248501798,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 8080
    },
    {
      "epoch": 3.2285257690771076,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 8081
    },
    {
      "epoch": 3.228925289652417,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 8082
    },
    {
      "epoch": 3.2293248102277268,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 8083
    },
    {
      "epoch": 3.2297243308030366,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 8084
    },
    {
      "epoch": 3.230123851378346,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 8085
    },
    {
      "epoch": 3.2305233719536557,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 8086
    },
    {
      "epoch": 3.230922892528965,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 8087
    },
    {
      "epoch": 3.231322413104275,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 8088
    },
    {
      "epoch": 3.2317219336795846,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 8089
    },
    {
      "epoch": 3.232121454254894,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 8090
    },
    {
      "epoch": 3.2325209748302037,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 8091
    },
    {
      "epoch": 3.2329204954055135,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0651,
      "step": 8092
    },
    {
      "epoch": 3.233320015980823,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 8093
    },
    {
      "epoch": 3.2337195365561326,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0669,
      "step": 8094
    },
    {
      "epoch": 3.2341190571314424,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 8095
    },
    {
      "epoch": 3.2345185777067518,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 8096
    },
    {
      "epoch": 3.2349180982820616,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 8097
    },
    {
      "epoch": 3.2353176188573713,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 8098
    },
    {
      "epoch": 3.2357171394326807,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 8099
    },
    {
      "epoch": 3.2361166600079905,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 8100
    },
    {
      "epoch": 3.2365161805833003,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 8101
    },
    {
      "epoch": 3.2369157011586096,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 8102
    },
    {
      "epoch": 3.2373152217339194,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 8103
    },
    {
      "epoch": 3.2377147423092287,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 8104
    },
    {
      "epoch": 3.2381142628845385,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 8105
    },
    {
      "epoch": 3.2385137834598483,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 8106
    },
    {
      "epoch": 3.2389133040351576,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 8107
    },
    {
      "epoch": 3.2393128246104674,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 8108
    },
    {
      "epoch": 3.239712345185777,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 8109
    },
    {
      "epoch": 3.2401118657610866,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 8110
    },
    {
      "epoch": 3.2405113863363963,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 8111
    },
    {
      "epoch": 3.240910906911706,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 8112
    },
    {
      "epoch": 3.2413104274870155,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 8113
    },
    {
      "epoch": 3.2417099480623253,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 8114
    },
    {
      "epoch": 3.2421094686376346,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 8115
    },
    {
      "epoch": 3.2425089892129444,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 8116
    },
    {
      "epoch": 3.242908509788254,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 8117
    },
    {
      "epoch": 3.2433080303635635,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 8118
    },
    {
      "epoch": 3.2437075509388733,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 8119
    },
    {
      "epoch": 3.244107071514183,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 8120
    },
    {
      "epoch": 3.2445065920894924,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 8121
    },
    {
      "epoch": 3.2449061126648022,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 8122
    },
    {
      "epoch": 3.245305633240112,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 8123
    },
    {
      "epoch": 3.2457051538154214,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 8124
    },
    {
      "epoch": 3.246104674390731,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 8125
    },
    {
      "epoch": 3.246504194966041,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 8126
    },
    {
      "epoch": 3.2469037155413503,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 8127
    },
    {
      "epoch": 3.24730323611666,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 8128
    },
    {
      "epoch": 3.24770275669197,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 8129
    },
    {
      "epoch": 3.248102277267279,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 8130
    },
    {
      "epoch": 3.248501797842589,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 8131
    },
    {
      "epoch": 3.2489013184178983,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 8132
    },
    {
      "epoch": 3.249300838993208,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 8133
    },
    {
      "epoch": 3.249700359568518,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 8134
    },
    {
      "epoch": 3.2500998801438272,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 8135
    },
    {
      "epoch": 3.250499400719137,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 8136
    },
    {
      "epoch": 3.250898921294447,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 8137
    },
    {
      "epoch": 3.251298441869756,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 8138
    },
    {
      "epoch": 3.251697962445066,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 8139
    },
    {
      "epoch": 3.2520974830203757,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 8140
    },
    {
      "epoch": 3.252497003595685,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 8141
    },
    {
      "epoch": 3.252896524170995,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 8142
    },
    {
      "epoch": 3.2532960447463046,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 8143
    },
    {
      "epoch": 3.253695565321614,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 8144
    },
    {
      "epoch": 3.2540950858969238,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 8145
    },
    {
      "epoch": 3.2544946064722335,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 8146
    },
    {
      "epoch": 3.254894127047543,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 8147
    },
    {
      "epoch": 3.2552936476228527,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 8148
    },
    {
      "epoch": 3.255693168198162,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 8149
    },
    {
      "epoch": 3.256092688773472,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 8150
    },
    {
      "epoch": 3.2564922093487816,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 8151
    },
    {
      "epoch": 3.256891729924091,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 8152
    },
    {
      "epoch": 3.2572912504994007,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 8153
    },
    {
      "epoch": 3.2576907710747105,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 8154
    },
    {
      "epoch": 3.25809029165002,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 8155
    },
    {
      "epoch": 3.2584898122253296,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 8156
    },
    {
      "epoch": 3.2588893328006394,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 8157
    },
    {
      "epoch": 3.2592888533759488,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 8158
    },
    {
      "epoch": 3.2596883739512585,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 8159
    },
    {
      "epoch": 3.260087894526568,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 8160
    },
    {
      "epoch": 3.2604874151018777,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 8161
    },
    {
      "epoch": 3.2608869356771875,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 8162
    },
    {
      "epoch": 3.2612864562524972,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 8163
    },
    {
      "epoch": 3.2616859768278066,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 8164
    },
    {
      "epoch": 3.2620854974031164,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 8165
    },
    {
      "epoch": 3.2624850179784257,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 8166
    },
    {
      "epoch": 3.2628845385537355,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 8167
    },
    {
      "epoch": 3.2632840591290453,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 8168
    },
    {
      "epoch": 3.2636835797043546,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 8169
    },
    {
      "epoch": 3.2640831002796644,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 8170
    },
    {
      "epoch": 3.264482620854974,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 8171
    },
    {
      "epoch": 3.2648821414302835,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 8172
    },
    {
      "epoch": 3.2652816620055933,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 8173
    },
    {
      "epoch": 3.265681182580903,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 8174
    },
    {
      "epoch": 3.2660807031562125,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 8175
    },
    {
      "epoch": 3.2664802237315222,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 8176
    },
    {
      "epoch": 3.2668797443068316,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 8177
    },
    {
      "epoch": 3.2672792648821414,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 8178
    },
    {
      "epoch": 3.267678785457451,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 8179
    },
    {
      "epoch": 3.2680783060327605,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 8180
    },
    {
      "epoch": 3.2684778266080703,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 8181
    },
    {
      "epoch": 3.26887734718338,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 8182
    },
    {
      "epoch": 3.2692768677586894,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 8183
    },
    {
      "epoch": 3.269676388333999,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 8184
    },
    {
      "epoch": 3.270075908909309,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 8185
    },
    {
      "epoch": 3.2704754294846183,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 8186
    },
    {
      "epoch": 3.270874950059928,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 8187
    },
    {
      "epoch": 3.271274470635238,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0675,
      "step": 8188
    },
    {
      "epoch": 3.2716739912105472,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 8189
    },
    {
      "epoch": 3.272073511785857,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 8190
    },
    {
      "epoch": 3.272473032361167,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 8191
    },
    {
      "epoch": 3.272872552936476,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 8192
    },
    {
      "epoch": 3.273272073511786,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 8193
    },
    {
      "epoch": 3.2736715940870953,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 8194
    },
    {
      "epoch": 3.274071114662405,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 8195
    },
    {
      "epoch": 3.274470635237715,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 8196
    },
    {
      "epoch": 3.274870155813024,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 8197
    },
    {
      "epoch": 3.275269676388334,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 8198
    },
    {
      "epoch": 3.275669196963644,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 8199
    },
    {
      "epoch": 3.276068717538953,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 8200
    },
    {
      "epoch": 3.276468238114263,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 8201
    },
    {
      "epoch": 3.2768677586895727,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 8202
    },
    {
      "epoch": 3.277267279264882,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 8203
    },
    {
      "epoch": 3.277666799840192,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 8204
    },
    {
      "epoch": 3.278066320415501,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 8205
    },
    {
      "epoch": 3.278465840990811,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 8206
    },
    {
      "epoch": 3.2788653615661207,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 8207
    },
    {
      "epoch": 3.2792648821414305,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 8208
    },
    {
      "epoch": 3.27966440271674,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 8209
    },
    {
      "epoch": 3.2800639232920497,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 8210
    },
    {
      "epoch": 3.280463443867359,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 8211
    },
    {
      "epoch": 3.280862964442669,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 8212
    },
    {
      "epoch": 3.2812624850179786,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 8213
    },
    {
      "epoch": 3.281662005593288,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 8214
    },
    {
      "epoch": 3.2820615261685977,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 8215
    },
    {
      "epoch": 3.2824610467439075,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 8216
    },
    {
      "epoch": 3.282860567319217,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 8217
    },
    {
      "epoch": 3.2832600878945266,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 8218
    },
    {
      "epoch": 3.2836596084698364,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 8219
    },
    {
      "epoch": 3.2840591290451457,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 8220
    },
    {
      "epoch": 3.2844586496204555,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 8221
    },
    {
      "epoch": 3.284858170195765,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 8222
    },
    {
      "epoch": 3.2852576907710747,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 8223
    },
    {
      "epoch": 3.2856572113463844,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 8224
    },
    {
      "epoch": 3.286056731921694,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 8225
    },
    {
      "epoch": 3.2864562524970036,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 8226
    },
    {
      "epoch": 3.2868557730723134,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 8227
    },
    {
      "epoch": 3.2872552936476227,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 8228
    },
    {
      "epoch": 3.2876548142229325,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 8229
    },
    {
      "epoch": 3.2880543347982423,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 8230
    },
    {
      "epoch": 3.2884538553735516,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 8231
    },
    {
      "epoch": 3.2888533759488614,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 8232
    },
    {
      "epoch": 3.289252896524171,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 8233
    },
    {
      "epoch": 3.2896524170994805,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 8234
    },
    {
      "epoch": 3.2900519376747903,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 8235
    },
    {
      "epoch": 3.2904514582501,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 8236
    },
    {
      "epoch": 3.2908509788254094,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 8237
    },
    {
      "epoch": 3.2912504994007192,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 8238
    },
    {
      "epoch": 3.2916500199760286,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 8239
    },
    {
      "epoch": 3.2920495405513384,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 8240
    },
    {
      "epoch": 3.292449061126648,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 8241
    },
    {
      "epoch": 3.2928485817019575,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 8242
    },
    {
      "epoch": 3.2932481022772673,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 8243
    },
    {
      "epoch": 3.293647622852577,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 8244
    },
    {
      "epoch": 3.2940471434278864,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 8245
    },
    {
      "epoch": 3.294446664003196,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 8246
    },
    {
      "epoch": 3.294846184578506,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 8247
    },
    {
      "epoch": 3.2952457051538153,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 8248
    },
    {
      "epoch": 3.295645225729125,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 8249
    },
    {
      "epoch": 3.2960447463044344,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 8250
    },
    {
      "epoch": 3.2964442668797442,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 8251
    },
    {
      "epoch": 3.296843787455054,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 8252
    },
    {
      "epoch": 3.297243308030364,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 8253
    },
    {
      "epoch": 3.297642828605673,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 8254
    },
    {
      "epoch": 3.298042349180983,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 8255
    },
    {
      "epoch": 3.2984418697562923,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 8256
    },
    {
      "epoch": 3.298841390331602,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 8257
    },
    {
      "epoch": 3.299240910906912,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 8258
    },
    {
      "epoch": 3.299640431482221,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 8259
    },
    {
      "epoch": 3.300039952057531,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 8260
    },
    {
      "epoch": 3.3004394726328408,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 8261
    },
    {
      "epoch": 3.30083899320815,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 8262
    },
    {
      "epoch": 3.30123851378346,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 8263
    },
    {
      "epoch": 3.3016380343587697,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 8264
    },
    {
      "epoch": 3.302037554934079,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 8265
    },
    {
      "epoch": 3.302437075509389,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 8266
    },
    {
      "epoch": 3.302836596084698,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 8267
    },
    {
      "epoch": 3.303236116660008,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 8268
    },
    {
      "epoch": 3.3036356372353177,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 8269
    },
    {
      "epoch": 3.304035157810627,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 8270
    },
    {
      "epoch": 3.304434678385937,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 8271
    },
    {
      "epoch": 3.3048341989612466,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 8272
    },
    {
      "epoch": 3.305233719536556,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 8273
    },
    {
      "epoch": 3.3056332401118658,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 8274
    },
    {
      "epoch": 3.3060327606871756,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 8275
    },
    {
      "epoch": 3.306432281262485,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 8276
    },
    {
      "epoch": 3.3068318018377947,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 8277
    },
    {
      "epoch": 3.3072313224131045,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 8278
    },
    {
      "epoch": 3.307630842988414,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0651,
      "step": 8279
    },
    {
      "epoch": 3.3080303635637236,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 8280
    },
    {
      "epoch": 3.3084298841390334,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 8281
    },
    {
      "epoch": 3.3088294047143427,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 8282
    },
    {
      "epoch": 3.3092289252896525,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 8283
    },
    {
      "epoch": 3.309628445864962,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 8284
    },
    {
      "epoch": 3.3100279664402716,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 8285
    },
    {
      "epoch": 3.3104274870155814,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 8286
    },
    {
      "epoch": 3.3108270075908908,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 8287
    },
    {
      "epoch": 3.3112265281662006,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 8288
    },
    {
      "epoch": 3.3116260487415103,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 8289
    },
    {
      "epoch": 3.3120255693168197,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 8290
    },
    {
      "epoch": 3.3124250898921295,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 8291
    },
    {
      "epoch": 3.3128246104674393,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 8292
    },
    {
      "epoch": 3.3132241310427486,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 8293
    },
    {
      "epoch": 3.3136236516180584,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 8294
    },
    {
      "epoch": 3.3140231721933677,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 8295
    },
    {
      "epoch": 3.3144226927686775,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 8296
    },
    {
      "epoch": 3.3148222133439873,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 8297
    },
    {
      "epoch": 3.315221733919297,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 8298
    },
    {
      "epoch": 3.3156212544946064,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 8299
    },
    {
      "epoch": 3.316020775069916,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 8300
    },
    {
      "epoch": 3.3164202956452256,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 8301
    },
    {
      "epoch": 3.3168198162205353,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 8302
    },
    {
      "epoch": 3.317219336795845,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 8303
    },
    {
      "epoch": 3.3176188573711545,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 8304
    },
    {
      "epoch": 3.3180183779464643,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 8305
    },
    {
      "epoch": 3.318417898521774,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 8306
    },
    {
      "epoch": 3.3188174190970834,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 8307
    },
    {
      "epoch": 3.319216939672393,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 8308
    },
    {
      "epoch": 3.319616460247703,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 8309
    },
    {
      "epoch": 3.3200159808230123,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 8310
    },
    {
      "epoch": 3.320415501398322,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 8311
    },
    {
      "epoch": 3.3208150219736314,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 8312
    },
    {
      "epoch": 3.321214542548941,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 8313
    },
    {
      "epoch": 3.321614063124251,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 8314
    },
    {
      "epoch": 3.3220135836995603,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 8315
    },
    {
      "epoch": 3.32241310427487,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 8316
    },
    {
      "epoch": 3.32281262485018,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 8317
    },
    {
      "epoch": 3.3232121454254893,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 8318
    },
    {
      "epoch": 3.323611666000799,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 8319
    },
    {
      "epoch": 3.324011186576109,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 8320
    },
    {
      "epoch": 3.324410707151418,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 8321
    },
    {
      "epoch": 3.324810227726728,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 8322
    },
    {
      "epoch": 3.3252097483020377,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 8323
    },
    {
      "epoch": 3.325609268877347,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 8324
    },
    {
      "epoch": 3.326008789452657,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 8325
    },
    {
      "epoch": 3.3264083100279667,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 8326
    },
    {
      "epoch": 3.326807830603276,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 8327
    },
    {
      "epoch": 3.327207351178586,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 8328
    },
    {
      "epoch": 3.327606871753895,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 8329
    },
    {
      "epoch": 3.328006392329205,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 8330
    },
    {
      "epoch": 3.3284059129045147,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 8331
    },
    {
      "epoch": 3.328805433479824,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 8332
    },
    {
      "epoch": 3.329204954055134,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 8333
    },
    {
      "epoch": 3.3296044746304436,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 8334
    },
    {
      "epoch": 3.330003995205753,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 8335
    },
    {
      "epoch": 3.3304035157810628,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 8336
    },
    {
      "epoch": 3.3308030363563725,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 8337
    },
    {
      "epoch": 3.331202556931682,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 8338
    },
    {
      "epoch": 3.3316020775069917,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 8339
    },
    {
      "epoch": 3.332001598082301,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 8340
    },
    {
      "epoch": 3.332401118657611,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 8341
    },
    {
      "epoch": 3.3328006392329206,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 8342
    },
    {
      "epoch": 3.3332001598082304,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 8343
    },
    {
      "epoch": 3.3335996803835397,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 8344
    },
    {
      "epoch": 3.3339992009588495,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 8345
    },
    {
      "epoch": 3.334398721534159,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 8346
    },
    {
      "epoch": 3.3347982421094686,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 8347
    },
    {
      "epoch": 3.3351977626847784,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 8348
    },
    {
      "epoch": 3.3355972832600878,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 8349
    },
    {
      "epoch": 3.3359968038353975,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 8350
    },
    {
      "epoch": 3.3363963244107073,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 8351
    },
    {
      "epoch": 3.3367958449860167,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 8352
    },
    {
      "epoch": 3.3371953655613265,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 8353
    },
    {
      "epoch": 3.3375948861366362,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 8354
    },
    {
      "epoch": 3.3379944067119456,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 8355
    },
    {
      "epoch": 3.3383939272872554,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 8356
    },
    {
      "epoch": 3.3387934478625647,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 8357
    },
    {
      "epoch": 3.3391929684378745,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 8358
    },
    {
      "epoch": 3.3395924890131843,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 8359
    },
    {
      "epoch": 3.3399920095884936,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 8360
    },
    {
      "epoch": 3.3403915301638034,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 8361
    },
    {
      "epoch": 3.340791050739113,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 8362
    },
    {
      "epoch": 3.3411905713144225,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 8363
    },
    {
      "epoch": 3.3415900918897323,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 8364
    },
    {
      "epoch": 3.341989612465042,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 8365
    },
    {
      "epoch": 3.3423891330403515,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 8366
    },
    {
      "epoch": 3.3427886536156612,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 8367
    },
    {
      "epoch": 3.343188174190971,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 8368
    },
    {
      "epoch": 3.3435876947662804,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 8369
    },
    {
      "epoch": 3.34398721534159,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 8370
    },
    {
      "epoch": 3.3443867359169,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 8371
    },
    {
      "epoch": 3.3447862564922093,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0668,
      "step": 8372
    },
    {
      "epoch": 3.345185777067519,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 8373
    },
    {
      "epoch": 3.3455852976428284,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 8374
    },
    {
      "epoch": 3.345984818218138,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 8375
    },
    {
      "epoch": 3.346384338793448,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 8376
    },
    {
      "epoch": 3.3467838593687573,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 8377
    },
    {
      "epoch": 3.347183379944067,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 8378
    },
    {
      "epoch": 3.347582900519377,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 8379
    },
    {
      "epoch": 3.3479824210946862,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 8380
    },
    {
      "epoch": 3.348381941669996,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 8381
    },
    {
      "epoch": 3.348781462245306,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 8382
    },
    {
      "epoch": 3.349180982820615,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 8383
    },
    {
      "epoch": 3.349580503395925,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 8384
    },
    {
      "epoch": 3.3499800239712343,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 8385
    },
    {
      "epoch": 3.350379544546544,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 8386
    },
    {
      "epoch": 3.350779065121854,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 8387
    },
    {
      "epoch": 3.3511785856971636,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 8388
    },
    {
      "epoch": 3.351578106272473,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 8389
    },
    {
      "epoch": 3.3519776268477828,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 8390
    },
    {
      "epoch": 3.352377147423092,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 8391
    },
    {
      "epoch": 3.352776667998402,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 8392
    },
    {
      "epoch": 3.3531761885737117,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 8393
    },
    {
      "epoch": 3.353575709149021,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 8394
    },
    {
      "epoch": 3.353975229724331,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 8395
    },
    {
      "epoch": 3.3543747502996406,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 8396
    },
    {
      "epoch": 3.35477427087495,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 8397
    },
    {
      "epoch": 3.3551737914502597,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 8398
    },
    {
      "epoch": 3.3555733120255695,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 8399
    },
    {
      "epoch": 3.355972832600879,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 8400
    },
    {
      "epoch": 3.3563723531761886,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 8401
    },
    {
      "epoch": 3.356771873751498,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 8402
    },
    {
      "epoch": 3.357171394326808,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 8403
    },
    {
      "epoch": 3.3575709149021176,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 8404
    },
    {
      "epoch": 3.357970435477427,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 8405
    },
    {
      "epoch": 3.3583699560527367,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 8406
    },
    {
      "epoch": 3.3587694766280465,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 8407
    },
    {
      "epoch": 3.359168997203356,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 8408
    },
    {
      "epoch": 3.3595685177786656,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 8409
    },
    {
      "epoch": 3.3599680383539754,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 8410
    },
    {
      "epoch": 3.3603675589292847,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 8411
    },
    {
      "epoch": 3.3607670795045945,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 8412
    },
    {
      "epoch": 3.361166600079904,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 8413
    },
    {
      "epoch": 3.3615661206552137,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 8414
    },
    {
      "epoch": 3.3619656412305234,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 8415
    },
    {
      "epoch": 3.3623651618058332,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 8416
    },
    {
      "epoch": 3.3627646823811426,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 8417
    },
    {
      "epoch": 3.3631642029564524,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 8418
    },
    {
      "epoch": 3.3635637235317617,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 8419
    },
    {
      "epoch": 3.3639632441070715,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 8420
    },
    {
      "epoch": 3.3643627646823813,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 8421
    },
    {
      "epoch": 3.3647622852576906,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 8422
    },
    {
      "epoch": 3.3651618058330004,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 8423
    },
    {
      "epoch": 3.36556132640831,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 8424
    },
    {
      "epoch": 3.3659608469836195,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 8425
    },
    {
      "epoch": 3.3663603675589293,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 8426
    },
    {
      "epoch": 3.366759888134239,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 8427
    },
    {
      "epoch": 3.3671594087095484,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 8428
    },
    {
      "epoch": 3.3675589292848582,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 8429
    },
    {
      "epoch": 3.3679584498601676,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 8430
    },
    {
      "epoch": 3.3683579704354774,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 8431
    },
    {
      "epoch": 3.368757491010787,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 8432
    },
    {
      "epoch": 3.369157011586097,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 8433
    },
    {
      "epoch": 3.3695565321614063,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 8434
    },
    {
      "epoch": 3.369956052736716,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 8435
    },
    {
      "epoch": 3.3703555733120254,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 8436
    },
    {
      "epoch": 3.370755093887335,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 8437
    },
    {
      "epoch": 3.371154614462645,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 8438
    },
    {
      "epoch": 3.3715541350379543,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 8439
    },
    {
      "epoch": 3.371953655613264,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 8440
    },
    {
      "epoch": 3.372353176188574,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 8441
    },
    {
      "epoch": 3.3727526967638832,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 8442
    },
    {
      "epoch": 3.373152217339193,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 8443
    },
    {
      "epoch": 3.373551737914503,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 8444
    },
    {
      "epoch": 3.373951258489812,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 8445
    },
    {
      "epoch": 3.374350779065122,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 8446
    },
    {
      "epoch": 3.3747502996404313,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 8447
    },
    {
      "epoch": 3.375149820215741,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 8448
    },
    {
      "epoch": 3.375549340791051,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 8449
    },
    {
      "epoch": 3.37594886136636,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 8450
    },
    {
      "epoch": 3.37634838194167,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 8451
    },
    {
      "epoch": 3.3767479025169798,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 8452
    },
    {
      "epoch": 3.377147423092289,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 8453
    },
    {
      "epoch": 3.377546943667599,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 8454
    },
    {
      "epoch": 3.3779464642429087,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 8455
    },
    {
      "epoch": 3.378345984818218,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 8456
    },
    {
      "epoch": 3.378745505393528,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 8457
    },
    {
      "epoch": 3.379145025968837,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 8458
    },
    {
      "epoch": 3.379544546544147,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 8459
    },
    {
      "epoch": 3.3799440671194567,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 8460
    },
    {
      "epoch": 3.3803435876947665,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 8461
    },
    {
      "epoch": 3.380743108270076,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 8462
    },
    {
      "epoch": 3.3811426288453856,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0698,
      "step": 8463
    },
    {
      "epoch": 3.381542149420695,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 8464
    },
    {
      "epoch": 3.3819416699960048,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 8465
    },
    {
      "epoch": 3.3823411905713145,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 8466
    },
    {
      "epoch": 3.382740711146624,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 8467
    },
    {
      "epoch": 3.3831402317219337,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 8468
    },
    {
      "epoch": 3.3835397522972435,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 8469
    },
    {
      "epoch": 3.383939272872553,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 8470
    },
    {
      "epoch": 3.3843387934478626,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 8471
    },
    {
      "epoch": 3.3847383140231724,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 8472
    },
    {
      "epoch": 3.3851378345984817,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 8473
    },
    {
      "epoch": 3.3855373551737915,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 8474
    },
    {
      "epoch": 3.385936875749101,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 8475
    },
    {
      "epoch": 3.3863363963244106,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 8476
    },
    {
      "epoch": 3.3867359168997204,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 8477
    },
    {
      "epoch": 3.38713543747503,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 8478
    },
    {
      "epoch": 3.3875349580503396,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 8479
    },
    {
      "epoch": 3.3879344786256493,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 8480
    },
    {
      "epoch": 3.3883339992009587,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 8481
    },
    {
      "epoch": 3.3887335197762685,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 8482
    },
    {
      "epoch": 3.3891330403515783,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 8483
    },
    {
      "epoch": 3.3895325609268876,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 8484
    },
    {
      "epoch": 3.3899320815021974,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 8485
    },
    {
      "epoch": 3.390331602077507,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 8486
    },
    {
      "epoch": 3.3907311226528165,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 8487
    },
    {
      "epoch": 3.3911306432281263,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 8488
    },
    {
      "epoch": 3.391530163803436,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 8489
    },
    {
      "epoch": 3.3919296843787454,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 8490
    },
    {
      "epoch": 3.392329204954055,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 8491
    },
    {
      "epoch": 3.3927287255293646,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 8492
    },
    {
      "epoch": 3.3931282461046743,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 8493
    },
    {
      "epoch": 3.393527766679984,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 8494
    },
    {
      "epoch": 3.3939272872552935,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 8495
    },
    {
      "epoch": 3.3943268078306033,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 8496
    },
    {
      "epoch": 3.394726328405913,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 8497
    },
    {
      "epoch": 3.3951258489812224,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 8498
    },
    {
      "epoch": 3.395525369556532,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 8499
    },
    {
      "epoch": 3.395924890131842,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 8500
    },
    {
      "epoch": 3.3963244107071513,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 8501
    },
    {
      "epoch": 3.396723931282461,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 8502
    },
    {
      "epoch": 3.3971234518577704,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 8503
    },
    {
      "epoch": 3.39752297243308,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 8504
    },
    {
      "epoch": 3.39792249300839,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 8505
    },
    {
      "epoch": 3.3983220135837,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 8506
    },
    {
      "epoch": 3.398721534159009,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 8507
    },
    {
      "epoch": 3.399121054734319,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 8508
    },
    {
      "epoch": 3.3995205753096283,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 8509
    },
    {
      "epoch": 3.399920095884938,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 8510
    },
    {
      "epoch": 3.400319616460248,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 8511
    },
    {
      "epoch": 3.400719137035557,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 8512
    },
    {
      "epoch": 3.401118657610867,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 8513
    },
    {
      "epoch": 3.4015181781861767,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 8514
    },
    {
      "epoch": 3.401917698761486,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 8515
    },
    {
      "epoch": 3.402317219336796,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 8516
    },
    {
      "epoch": 3.4027167399121057,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 8517
    },
    {
      "epoch": 3.403116260487415,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 8518
    },
    {
      "epoch": 3.403515781062725,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 8519
    },
    {
      "epoch": 3.403915301638034,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 8520
    },
    {
      "epoch": 3.404314822213344,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 8521
    },
    {
      "epoch": 3.4047143427886537,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 8522
    },
    {
      "epoch": 3.4051138633639635,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 8523
    },
    {
      "epoch": 3.405513383939273,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 8524
    },
    {
      "epoch": 3.4059129045145826,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 8525
    },
    {
      "epoch": 3.406312425089892,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 8526
    },
    {
      "epoch": 3.4067119456652017,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 8527
    },
    {
      "epoch": 3.4071114662405115,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 8528
    },
    {
      "epoch": 3.407510986815821,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 8529
    },
    {
      "epoch": 3.4079105073911307,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 8530
    },
    {
      "epoch": 3.4083100279664404,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 8531
    },
    {
      "epoch": 3.40870954854175,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 8532
    },
    {
      "epoch": 3.4091090691170596,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 8533
    },
    {
      "epoch": 3.4095085896923694,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 8534
    },
    {
      "epoch": 3.4099081102676787,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 8535
    },
    {
      "epoch": 3.4103076308429885,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 8536
    },
    {
      "epoch": 3.410707151418298,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 8537
    },
    {
      "epoch": 3.4111066719936076,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 8538
    },
    {
      "epoch": 3.4115061925689174,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 8539
    },
    {
      "epoch": 3.4119057131442267,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 8540
    },
    {
      "epoch": 3.4123052337195365,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 8541
    },
    {
      "epoch": 3.4127047542948463,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 8542
    },
    {
      "epoch": 3.4131042748701557,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 8543
    },
    {
      "epoch": 3.4135037954454654,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 8544
    },
    {
      "epoch": 3.4139033160207752,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 8545
    },
    {
      "epoch": 3.4143028365960846,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 8546
    },
    {
      "epoch": 3.4147023571713944,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 8547
    },
    {
      "epoch": 3.4151018777467037,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 8548
    },
    {
      "epoch": 3.4155013983220135,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 8549
    },
    {
      "epoch": 3.4159009188973233,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 8550
    },
    {
      "epoch": 3.416300439472633,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 8551
    },
    {
      "epoch": 3.4166999600479424,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 8552
    },
    {
      "epoch": 3.417099480623252,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 8553
    },
    {
      "epoch": 3.4174990011985615,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 8554
    },
    {
      "epoch": 3.4178985217738713,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 8555
    },
    {
      "epoch": 3.418298042349181,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 8556
    },
    {
      "epoch": 3.4186975629244905,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 8557
    },
    {
      "epoch": 3.4190970834998002,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 8558
    },
    {
      "epoch": 3.41949660407511,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 8559
    },
    {
      "epoch": 3.4198961246504194,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 8560
    },
    {
      "epoch": 3.420295645225729,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 8561
    },
    {
      "epoch": 3.420695165801039,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 8562
    },
    {
      "epoch": 3.4210946863763483,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 8563
    },
    {
      "epoch": 3.421494206951658,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 8564
    },
    {
      "epoch": 3.4218937275269674,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 8565
    },
    {
      "epoch": 3.422293248102277,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 8566
    },
    {
      "epoch": 3.422692768677587,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 8567
    },
    {
      "epoch": 3.4230922892528968,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 8568
    },
    {
      "epoch": 3.423491809828206,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 8569
    },
    {
      "epoch": 3.423891330403516,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 8570
    },
    {
      "epoch": 3.4242908509788252,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 8571
    },
    {
      "epoch": 3.424690371554135,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 8572
    },
    {
      "epoch": 3.425089892129445,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 8573
    },
    {
      "epoch": 3.425489412704754,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 8574
    },
    {
      "epoch": 3.425888933280064,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 8575
    },
    {
      "epoch": 3.4262884538553737,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 8576
    },
    {
      "epoch": 3.426687974430683,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 8577
    },
    {
      "epoch": 3.427087495005993,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 8578
    },
    {
      "epoch": 3.4274870155813026,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 8579
    },
    {
      "epoch": 3.427886536156612,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 8580
    },
    {
      "epoch": 3.4282860567319218,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 8581
    },
    {
      "epoch": 3.428685577307231,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 8582
    },
    {
      "epoch": 3.429085097882541,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 8583
    },
    {
      "epoch": 3.4294846184578507,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 8584
    },
    {
      "epoch": 3.42988413903316,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 8585
    },
    {
      "epoch": 3.43028365960847,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 8586
    },
    {
      "epoch": 3.4306831801837796,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 8587
    },
    {
      "epoch": 3.431082700759089,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 8588
    },
    {
      "epoch": 3.4314822213343987,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 8589
    },
    {
      "epoch": 3.4318817419097085,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 8590
    },
    {
      "epoch": 3.432281262485018,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 8591
    },
    {
      "epoch": 3.4326807830603276,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0709,
      "step": 8592
    },
    {
      "epoch": 3.433080303635637,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 8593
    },
    {
      "epoch": 3.4334798242109468,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 8594
    },
    {
      "epoch": 3.4338793447862566,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 8595
    },
    {
      "epoch": 3.4342788653615663,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 8596
    },
    {
      "epoch": 3.4346783859368757,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 8597
    },
    {
      "epoch": 3.4350779065121855,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 8598
    },
    {
      "epoch": 3.435477427087495,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 8599
    },
    {
      "epoch": 3.4358769476628046,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 8600
    },
    {
      "epoch": 3.4362764682381144,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 8601
    },
    {
      "epoch": 3.4366759888134237,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 8602
    },
    {
      "epoch": 3.4370755093887335,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 8603
    },
    {
      "epoch": 3.4374750299640433,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 8604
    },
    {
      "epoch": 3.4378745505393526,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0691,
      "step": 8605
    },
    {
      "epoch": 3.4382740711146624,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 8606
    },
    {
      "epoch": 3.438673591689972,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 8607
    },
    {
      "epoch": 3.4390731122652816,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 8608
    },
    {
      "epoch": 3.4394726328405913,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 8609
    },
    {
      "epoch": 3.4398721534159007,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 8610
    },
    {
      "epoch": 3.4402716739912105,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 8611
    },
    {
      "epoch": 3.4406711945665203,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 8612
    },
    {
      "epoch": 3.44107071514183,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 8613
    },
    {
      "epoch": 3.4414702357171394,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 8614
    },
    {
      "epoch": 3.441869756292449,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 8615
    },
    {
      "epoch": 3.4422692768677585,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 8616
    },
    {
      "epoch": 3.4426687974430683,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 8617
    },
    {
      "epoch": 3.443068318018378,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 8618
    },
    {
      "epoch": 3.4434678385936874,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 8619
    },
    {
      "epoch": 3.443867359168997,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 8620
    },
    {
      "epoch": 3.444266879744307,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 8621
    },
    {
      "epoch": 3.4446664003196164,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 8622
    },
    {
      "epoch": 3.445065920894926,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 8623
    },
    {
      "epoch": 3.445465441470236,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 8624
    },
    {
      "epoch": 3.4458649620455453,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 8625
    },
    {
      "epoch": 3.446264482620855,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 8626
    },
    {
      "epoch": 3.4466640031961644,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 8627
    },
    {
      "epoch": 3.447063523771474,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 8628
    },
    {
      "epoch": 3.447463044346784,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 8629
    },
    {
      "epoch": 3.4478625649220933,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 8630
    },
    {
      "epoch": 3.448262085497403,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 8631
    },
    {
      "epoch": 3.448661606072713,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 8632
    },
    {
      "epoch": 3.4490611266480222,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 8633
    },
    {
      "epoch": 3.449460647223332,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 8634
    },
    {
      "epoch": 3.449860167798642,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 8635
    },
    {
      "epoch": 3.450259688373951,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 8636
    },
    {
      "epoch": 3.450659208949261,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 8637
    },
    {
      "epoch": 3.4510587295245703,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 8638
    },
    {
      "epoch": 3.45145825009988,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 8639
    },
    {
      "epoch": 3.45185777067519,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 8640
    },
    {
      "epoch": 3.4522572912504996,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 8641
    },
    {
      "epoch": 3.452656811825809,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 8642
    },
    {
      "epoch": 3.4530563324011188,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 8643
    },
    {
      "epoch": 3.453455852976428,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 8644
    },
    {
      "epoch": 3.453855373551738,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 8645
    },
    {
      "epoch": 3.4542548941270477,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 8646
    },
    {
      "epoch": 3.454654414702357,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 8647
    },
    {
      "epoch": 3.455053935277667,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 8648
    },
    {
      "epoch": 3.4554534558529766,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.068,
      "step": 8649
    },
    {
      "epoch": 3.455852976428286,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 8650
    },
    {
      "epoch": 3.4562524970035957,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 8651
    },
    {
      "epoch": 3.4566520175789055,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 8652
    },
    {
      "epoch": 3.457051538154215,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 8653
    },
    {
      "epoch": 3.4574510587295246,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 8654
    },
    {
      "epoch": 3.457850579304834,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 8655
    },
    {
      "epoch": 3.4582500998801438,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 8656
    },
    {
      "epoch": 3.4586496204554535,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 8657
    },
    {
      "epoch": 3.4590491410307633,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 8658
    },
    {
      "epoch": 3.4594486616060727,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 8659
    },
    {
      "epoch": 3.4598481821813825,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 8660
    },
    {
      "epoch": 3.460247702756692,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 8661
    },
    {
      "epoch": 3.4606472233320016,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 8662
    },
    {
      "epoch": 3.4610467439073114,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 8663
    },
    {
      "epoch": 3.4614462644826207,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 8664
    },
    {
      "epoch": 3.4618457850579305,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 8665
    },
    {
      "epoch": 3.4622453056332403,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 8666
    },
    {
      "epoch": 3.4626448262085496,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 8667
    },
    {
      "epoch": 3.4630443467838594,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 8668
    },
    {
      "epoch": 3.463443867359169,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 8669
    },
    {
      "epoch": 3.4638433879344785,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 8670
    },
    {
      "epoch": 3.4642429085097883,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 8671
    },
    {
      "epoch": 3.4646424290850977,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 8672
    },
    {
      "epoch": 3.4650419496604075,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 8673
    },
    {
      "epoch": 3.4654414702357172,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 8674
    },
    {
      "epoch": 3.4658409908110266,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 8675
    },
    {
      "epoch": 3.4662405113863364,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 8676
    },
    {
      "epoch": 3.466640031961646,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 8677
    },
    {
      "epoch": 3.4670395525369555,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 8678
    },
    {
      "epoch": 3.4674390731122653,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 8679
    },
    {
      "epoch": 3.467838593687575,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 8680
    },
    {
      "epoch": 3.4682381142628844,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 8681
    },
    {
      "epoch": 3.468637634838194,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 8682
    },
    {
      "epoch": 3.4690371554135035,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 8683
    },
    {
      "epoch": 3.4694366759888133,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 8684
    },
    {
      "epoch": 3.469836196564123,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 8685
    },
    {
      "epoch": 3.470235717139433,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 8686
    },
    {
      "epoch": 3.4706352377147422,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 8687
    },
    {
      "epoch": 3.471034758290052,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 8688
    },
    {
      "epoch": 3.4714342788653614,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 8689
    },
    {
      "epoch": 3.471833799440671,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 8690
    },
    {
      "epoch": 3.472233320015981,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 8691
    },
    {
      "epoch": 3.4726328405912903,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 8692
    },
    {
      "epoch": 3.4730323611666,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 8693
    },
    {
      "epoch": 3.47343188174191,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 8694
    },
    {
      "epoch": 3.473831402317219,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 8695
    },
    {
      "epoch": 3.474230922892529,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 8696
    },
    {
      "epoch": 3.474630443467839,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 8697
    },
    {
      "epoch": 3.475029964043148,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 8698
    },
    {
      "epoch": 3.475429484618458,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 8699
    },
    {
      "epoch": 3.4758290051937673,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 8700
    },
    {
      "epoch": 3.476228525769077,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 8701
    },
    {
      "epoch": 3.476628046344387,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 8702
    },
    {
      "epoch": 3.4770275669196966,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 8703
    },
    {
      "epoch": 3.477427087495006,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 8704
    },
    {
      "epoch": 3.4778266080703157,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 8705
    },
    {
      "epoch": 3.478226128645625,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 8706
    },
    {
      "epoch": 3.478625649220935,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 8707
    },
    {
      "epoch": 3.4790251697962447,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 8708
    },
    {
      "epoch": 3.479424690371554,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 8709
    },
    {
      "epoch": 3.479824210946864,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 8710
    },
    {
      "epoch": 3.4802237315221736,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 8711
    },
    {
      "epoch": 3.480623252097483,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 8712
    },
    {
      "epoch": 3.4810227726727927,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 8713
    },
    {
      "epoch": 3.4814222932481025,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 8714
    },
    {
      "epoch": 3.481821813823412,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 8715
    },
    {
      "epoch": 3.4822213343987216,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 8716
    },
    {
      "epoch": 3.482620854974031,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 8717
    },
    {
      "epoch": 3.4830203755493407,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 8718
    },
    {
      "epoch": 3.4834198961246505,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 8719
    },
    {
      "epoch": 3.48381941669996,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.072,
      "step": 8720
    },
    {
      "epoch": 3.4842189372752697,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 8721
    },
    {
      "epoch": 3.4846184578505794,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 8722
    },
    {
      "epoch": 3.485017978425889,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 8723
    },
    {
      "epoch": 3.4854174990011986,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 8724
    },
    {
      "epoch": 3.4858170195765084,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 8725
    },
    {
      "epoch": 3.4862165401518177,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 8726
    },
    {
      "epoch": 3.4866160607271275,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 8727
    },
    {
      "epoch": 3.487015581302437,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 8728
    },
    {
      "epoch": 3.4874151018777466,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 8729
    },
    {
      "epoch": 3.4878146224530564,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 8730
    },
    {
      "epoch": 3.488214143028366,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 8731
    },
    {
      "epoch": 3.4886136636036755,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 8732
    },
    {
      "epoch": 3.4890131841789853,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 8733
    },
    {
      "epoch": 3.4894127047542947,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 8734
    },
    {
      "epoch": 3.4898122253296044,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 8735
    },
    {
      "epoch": 3.4902117459049142,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 8736
    },
    {
      "epoch": 3.4906112664802236,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 8737
    },
    {
      "epoch": 3.4910107870555334,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 8738
    },
    {
      "epoch": 3.491410307630843,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 8739
    },
    {
      "epoch": 3.4918098282061525,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 8740
    },
    {
      "epoch": 3.4922093487814623,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 8741
    },
    {
      "epoch": 3.492608869356772,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 8742
    },
    {
      "epoch": 3.4930083899320814,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 8743
    },
    {
      "epoch": 3.493407910507391,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 8744
    },
    {
      "epoch": 3.4938074310827005,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 8745
    },
    {
      "epoch": 3.4942069516580103,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 8746
    },
    {
      "epoch": 3.49460647223332,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 8747
    },
    {
      "epoch": 3.49500599280863,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 8748
    },
    {
      "epoch": 3.4954055133839392,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 8749
    },
    {
      "epoch": 3.495805033959249,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 8750
    },
    {
      "epoch": 3.4962045545345584,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 8751
    },
    {
      "epoch": 3.496604075109868,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 8752
    },
    {
      "epoch": 3.497003595685178,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 8753
    },
    {
      "epoch": 3.4974031162604873,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 8754
    },
    {
      "epoch": 3.497802636835797,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 8755
    },
    {
      "epoch": 3.498202157411107,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 8756
    },
    {
      "epoch": 3.498601677986416,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 8757
    },
    {
      "epoch": 3.499001198561726,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 8758
    },
    {
      "epoch": 3.4994007191370358,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 8759
    },
    {
      "epoch": 3.499800239712345,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 8760
    },
    {
      "epoch": 3.500199760287655,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 8761
    },
    {
      "epoch": 3.5005992808629642,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 8762
    },
    {
      "epoch": 3.500998801438274,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 8763
    },
    {
      "epoch": 3.501398322013584,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 8764
    },
    {
      "epoch": 3.5017978425888936,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 8765
    },
    {
      "epoch": 3.502197363164203,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 8766
    },
    {
      "epoch": 3.5025968837395127,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 8767
    },
    {
      "epoch": 3.502996404314822,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 8768
    },
    {
      "epoch": 3.503395924890132,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 8769
    },
    {
      "epoch": 3.5037954454654416,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 8770
    },
    {
      "epoch": 3.504194966040751,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 8771
    },
    {
      "epoch": 3.5045944866160608,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 8772
    },
    {
      "epoch": 3.50499400719137,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 8773
    },
    {
      "epoch": 3.50539352776668,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 8774
    },
    {
      "epoch": 3.5057930483419897,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 8775
    },
    {
      "epoch": 3.5061925689172995,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 8776
    },
    {
      "epoch": 3.506592089492609,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 8777
    },
    {
      "epoch": 3.5069916100679186,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 8778
    },
    {
      "epoch": 3.507391130643228,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 8779
    },
    {
      "epoch": 3.5077906512185377,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 8780
    },
    {
      "epoch": 3.5081901717938475,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 8781
    },
    {
      "epoch": 3.508589692369157,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 8782
    },
    {
      "epoch": 3.5089892129444666,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 8783
    },
    {
      "epoch": 3.5093887335197764,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 8784
    },
    {
      "epoch": 3.5097882540950858,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 8785
    },
    {
      "epoch": 3.5101877746703956,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 8786
    },
    {
      "epoch": 3.5105872952457053,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 8787
    },
    {
      "epoch": 3.5109868158210147,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 8788
    },
    {
      "epoch": 3.5113863363963245,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 8789
    },
    {
      "epoch": 3.511785856971634,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 8790
    },
    {
      "epoch": 3.5121853775469436,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 8791
    },
    {
      "epoch": 3.5125848981222534,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 8792
    },
    {
      "epoch": 3.512984418697563,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 8793
    },
    {
      "epoch": 3.5133839392728725,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 8794
    },
    {
      "epoch": 3.5137834598481823,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 8795
    },
    {
      "epoch": 3.5141829804234916,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 8796
    },
    {
      "epoch": 3.5145825009988014,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 8797
    },
    {
      "epoch": 3.514982021574111,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 8798
    },
    {
      "epoch": 3.5153815421494206,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 8799
    },
    {
      "epoch": 3.5157810627247303,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 8800
    },
    {
      "epoch": 3.5161805833000397,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 8801
    },
    {
      "epoch": 3.5165801038753495,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 8802
    },
    {
      "epoch": 3.5169796244506593,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 8803
    },
    {
      "epoch": 3.517379145025969,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 8804
    },
    {
      "epoch": 3.5177786656012784,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 8805
    },
    {
      "epoch": 3.518178186176588,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 8806
    },
    {
      "epoch": 3.5185777067518975,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 8807
    },
    {
      "epoch": 3.5189772273272073,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 8808
    },
    {
      "epoch": 3.519376747902517,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 8809
    },
    {
      "epoch": 3.519776268477827,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 8810
    },
    {
      "epoch": 3.520175789053136,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 8811
    },
    {
      "epoch": 3.520575309628446,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 8812
    },
    {
      "epoch": 3.5209748302037553,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 8813
    },
    {
      "epoch": 3.521374350779065,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 8814
    },
    {
      "epoch": 3.521773871354375,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 8815
    },
    {
      "epoch": 3.5221733919296843,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 8816
    },
    {
      "epoch": 3.522572912504994,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 8817
    },
    {
      "epoch": 3.5229724330803034,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 8818
    },
    {
      "epoch": 3.523371953655613,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 8819
    },
    {
      "epoch": 3.523771474230923,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 8820
    },
    {
      "epoch": 3.5241709948062327,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 8821
    },
    {
      "epoch": 3.524570515381542,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 8822
    },
    {
      "epoch": 3.524970035956852,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 8823
    },
    {
      "epoch": 3.525369556532161,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 8824
    },
    {
      "epoch": 3.525769077107471,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 8825
    },
    {
      "epoch": 3.526168597682781,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 8826
    },
    {
      "epoch": 3.52656811825809,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 8827
    },
    {
      "epoch": 3.5269676388334,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 8828
    },
    {
      "epoch": 3.5273671594087097,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 8829
    },
    {
      "epoch": 3.527766679984019,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 8830
    },
    {
      "epoch": 3.528166200559329,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 8831
    },
    {
      "epoch": 3.5285657211346386,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 8832
    },
    {
      "epoch": 3.528965241709948,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 8833
    },
    {
      "epoch": 3.5293647622852578,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 8834
    },
    {
      "epoch": 3.529764282860567,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 8835
    },
    {
      "epoch": 3.530163803435877,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 8836
    },
    {
      "epoch": 3.5305633240111867,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 8837
    },
    {
      "epoch": 3.5309628445864965,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 8838
    },
    {
      "epoch": 3.531362365161806,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 8839
    },
    {
      "epoch": 3.5317618857371156,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 8840
    },
    {
      "epoch": 3.532161406312425,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 8841
    },
    {
      "epoch": 3.5325609268877347,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 8842
    },
    {
      "epoch": 3.5329604474630445,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 8843
    },
    {
      "epoch": 3.533359968038354,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 8844
    },
    {
      "epoch": 3.5337594886136636,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 8845
    },
    {
      "epoch": 3.534159009188973,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 8846
    },
    {
      "epoch": 3.5345585297642828,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 8847
    },
    {
      "epoch": 3.5349580503395925,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 8848
    },
    {
      "epoch": 3.5353575709149023,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 8849
    },
    {
      "epoch": 3.5357570914902117,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 8850
    },
    {
      "epoch": 3.5361566120655215,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 8851
    },
    {
      "epoch": 3.536556132640831,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 8852
    },
    {
      "epoch": 3.5369556532161406,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 8853
    },
    {
      "epoch": 3.5373551737914504,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 8854
    },
    {
      "epoch": 3.53775469436676,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 8855
    },
    {
      "epoch": 3.5381542149420695,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 8856
    },
    {
      "epoch": 3.5385537355173793,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 8857
    },
    {
      "epoch": 3.5389532560926886,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 8858
    },
    {
      "epoch": 3.5393527766679984,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 8859
    },
    {
      "epoch": 3.539752297243308,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 8860
    },
    {
      "epoch": 3.5401518178186175,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 8861
    },
    {
      "epoch": 3.5405513383939273,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 8862
    },
    {
      "epoch": 3.5409508589692367,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 8863
    },
    {
      "epoch": 3.5413503795445465,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 8864
    },
    {
      "epoch": 3.5417499001198562,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 8865
    },
    {
      "epoch": 3.542149420695166,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 8866
    },
    {
      "epoch": 3.5425489412704754,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 8867
    },
    {
      "epoch": 3.542948461845785,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 8868
    },
    {
      "epoch": 3.5433479824210945,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 8869
    },
    {
      "epoch": 3.5437475029964043,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 8870
    },
    {
      "epoch": 3.544147023571714,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 8871
    },
    {
      "epoch": 3.5445465441470234,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 8872
    },
    {
      "epoch": 3.544946064722333,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 8873
    },
    {
      "epoch": 3.545345585297643,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 8874
    },
    {
      "epoch": 3.5457451058729523,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 8875
    },
    {
      "epoch": 3.546144626448262,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 8876
    },
    {
      "epoch": 3.546544147023572,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 8877
    },
    {
      "epoch": 3.5469436675988812,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 8878
    },
    {
      "epoch": 3.547343188174191,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 8879
    },
    {
      "epoch": 3.5477427087495004,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 8880
    },
    {
      "epoch": 3.54814222932481,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 8881
    },
    {
      "epoch": 3.54854174990012,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 8882
    },
    {
      "epoch": 3.5489412704754297,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 8883
    },
    {
      "epoch": 3.549340791050739,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 8884
    },
    {
      "epoch": 3.549740311626049,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 8885
    },
    {
      "epoch": 3.550139832201358,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 8886
    },
    {
      "epoch": 3.550539352776668,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 8887
    },
    {
      "epoch": 3.5509388733519778,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 8888
    },
    {
      "epoch": 3.551338393927287,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 8889
    },
    {
      "epoch": 3.551737914502597,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 8890
    },
    {
      "epoch": 3.5521374350779062,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 8891
    },
    {
      "epoch": 3.552536955653216,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 8892
    },
    {
      "epoch": 3.552936476228526,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 8893
    },
    {
      "epoch": 3.5533359968038356,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 8894
    },
    {
      "epoch": 3.553735517379145,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 8895
    },
    {
      "epoch": 3.5541350379544547,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 8896
    },
    {
      "epoch": 3.554534558529764,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 8897
    },
    {
      "epoch": 3.554934079105074,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 8898
    },
    {
      "epoch": 3.5553335996803836,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 8899
    },
    {
      "epoch": 3.5557331202556934,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0651,
      "step": 8900
    },
    {
      "epoch": 3.556132640831003,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 8901
    },
    {
      "epoch": 3.5565321614063126,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 8902
    },
    {
      "epoch": 3.556931681981622,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 8903
    },
    {
      "epoch": 3.5573312025569317,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 8904
    },
    {
      "epoch": 3.5577307231322415,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 8905
    },
    {
      "epoch": 3.558130243707551,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 8906
    },
    {
      "epoch": 3.5585297642828606,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 8907
    },
    {
      "epoch": 3.55892928485817,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 8908
    },
    {
      "epoch": 3.5593288054334797,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 8909
    },
    {
      "epoch": 3.5597283260087895,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 8910
    },
    {
      "epoch": 3.5601278465840993,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 8911
    },
    {
      "epoch": 3.5605273671594087,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 8912
    },
    {
      "epoch": 3.5609268877347184,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 8913
    },
    {
      "epoch": 3.561326408310028,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 8914
    },
    {
      "epoch": 3.5617259288853376,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 8915
    },
    {
      "epoch": 3.5621254494606474,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 8916
    },
    {
      "epoch": 3.5625249700359567,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 8917
    },
    {
      "epoch": 3.5629244906112665,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 8918
    },
    {
      "epoch": 3.5633240111865763,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 8919
    },
    {
      "epoch": 3.5637235317618856,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 8920
    },
    {
      "epoch": 3.5641230523371954,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 8921
    },
    {
      "epoch": 3.564522572912505,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 8922
    },
    {
      "epoch": 3.5649220934878145,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 8923
    },
    {
      "epoch": 3.5653216140631243,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 8924
    },
    {
      "epoch": 3.5657211346384337,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0684,
      "step": 8925
    },
    {
      "epoch": 3.5661206552137434,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 8926
    },
    {
      "epoch": 3.5665201757890532,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 8927
    },
    {
      "epoch": 3.566919696364363,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 8928
    },
    {
      "epoch": 3.5673192169396724,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 8929
    },
    {
      "epoch": 3.567718737514982,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 8930
    },
    {
      "epoch": 3.5681182580902915,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 8931
    },
    {
      "epoch": 3.5685177786656013,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 8932
    },
    {
      "epoch": 3.568917299240911,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 8933
    },
    {
      "epoch": 3.5693168198162204,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 8934
    },
    {
      "epoch": 3.56971634039153,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 8935
    },
    {
      "epoch": 3.5701158609668395,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 8936
    },
    {
      "epoch": 3.5705153815421493,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 8937
    },
    {
      "epoch": 3.570914902117459,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 8938
    },
    {
      "epoch": 3.571314422692769,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 8939
    },
    {
      "epoch": 3.5717139432680782,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 8940
    },
    {
      "epoch": 3.572113463843388,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 8941
    },
    {
      "epoch": 3.5725129844186974,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 8942
    },
    {
      "epoch": 3.572912504994007,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 8943
    },
    {
      "epoch": 3.573312025569317,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 8944
    },
    {
      "epoch": 3.5737115461446267,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 8945
    },
    {
      "epoch": 3.574111066719936,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 8946
    },
    {
      "epoch": 3.574510587295246,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 8947
    },
    {
      "epoch": 3.574910107870555,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 8948
    },
    {
      "epoch": 3.575309628445865,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 8949
    },
    {
      "epoch": 3.5757091490211748,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 8950
    },
    {
      "epoch": 3.576108669596484,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 8951
    },
    {
      "epoch": 3.576508190171794,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 8952
    },
    {
      "epoch": 3.5769077107471032,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 8953
    },
    {
      "epoch": 3.577307231322413,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 8954
    },
    {
      "epoch": 3.577706751897723,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 8955
    },
    {
      "epoch": 3.5781062724730326,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 8956
    },
    {
      "epoch": 3.578505793048342,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 8957
    },
    {
      "epoch": 3.5789053136236517,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 8958
    },
    {
      "epoch": 3.579304834198961,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 8959
    },
    {
      "epoch": 3.579704354774271,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 8960
    },
    {
      "epoch": 3.5801038753495806,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 8961
    },
    {
      "epoch": 3.58050339592489,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 8962
    },
    {
      "epoch": 3.5809029165001998,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 8963
    },
    {
      "epoch": 3.5813024370755095,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 8964
    },
    {
      "epoch": 3.581701957650819,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 8965
    },
    {
      "epoch": 3.5821014782261287,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 8966
    },
    {
      "epoch": 3.5825009988014385,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 8967
    },
    {
      "epoch": 3.582900519376748,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 8968
    },
    {
      "epoch": 3.5833000399520576,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 8969
    },
    {
      "epoch": 3.583699560527367,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 8970
    },
    {
      "epoch": 3.5840990811026767,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 8971
    },
    {
      "epoch": 3.5844986016779865,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 8972
    },
    {
      "epoch": 3.5848981222532963,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 8973
    },
    {
      "epoch": 3.5852976428286056,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 8974
    },
    {
      "epoch": 3.5856971634039154,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 8975
    },
    {
      "epoch": 3.5860966839792248,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 8976
    },
    {
      "epoch": 3.5864962045545346,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 8977
    },
    {
      "epoch": 3.5868957251298443,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 8978
    },
    {
      "epoch": 3.5872952457051537,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 8979
    },
    {
      "epoch": 3.5876947662804635,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 8980
    },
    {
      "epoch": 3.588094286855773,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 8981
    },
    {
      "epoch": 3.5884938074310826,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 8982
    },
    {
      "epoch": 3.5888933280063924,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 8983
    },
    {
      "epoch": 3.589292848581702,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 8984
    },
    {
      "epoch": 3.5896923691570115,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 8985
    },
    {
      "epoch": 3.5900918897323213,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 8986
    },
    {
      "epoch": 3.5904914103076306,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 8987
    },
    {
      "epoch": 3.5908909308829404,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 8988
    },
    {
      "epoch": 3.59129045145825,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 8989
    },
    {
      "epoch": 3.59168997203356,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 8990
    },
    {
      "epoch": 3.5920894926088693,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 8991
    },
    {
      "epoch": 3.592489013184179,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 8992
    },
    {
      "epoch": 3.5928885337594885,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 8993
    },
    {
      "epoch": 3.5932880543347983,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 8994
    },
    {
      "epoch": 3.593687574910108,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 8995
    },
    {
      "epoch": 3.5940870954854174,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 8996
    },
    {
      "epoch": 3.594486616060727,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 8997
    },
    {
      "epoch": 3.5948861366360365,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 8998
    },
    {
      "epoch": 3.5952856572113463,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 8999
    },
    {
      "epoch": 3.595685177786656,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 9000
    },
    {
      "epoch": 3.596084698361966,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 9001
    },
    {
      "epoch": 3.596484218937275,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 9002
    },
    {
      "epoch": 3.596883739512585,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 9003
    },
    {
      "epoch": 3.5972832600878943,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 9004
    },
    {
      "epoch": 3.597682780663204,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 9005
    },
    {
      "epoch": 3.598082301238514,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 9006
    },
    {
      "epoch": 3.5984818218138233,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 9007
    },
    {
      "epoch": 3.598881342389133,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 9008
    },
    {
      "epoch": 3.599280862964443,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 9009
    },
    {
      "epoch": 3.599680383539752,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 9010
    },
    {
      "epoch": 3.600079904115062,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 9011
    },
    {
      "epoch": 3.6004794246903717,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 9012
    },
    {
      "epoch": 3.600878945265681,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 9013
    },
    {
      "epoch": 3.601278465840991,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 9014
    },
    {
      "epoch": 3.6016779864163,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 9015
    },
    {
      "epoch": 3.60207750699161,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 9016
    },
    {
      "epoch": 3.60247702756692,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 9017
    },
    {
      "epoch": 3.6028765481422296,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 9018
    },
    {
      "epoch": 3.603276068717539,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 9019
    },
    {
      "epoch": 3.6036755892928487,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 9020
    },
    {
      "epoch": 3.604075109868158,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 9021
    },
    {
      "epoch": 3.604474630443468,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 9022
    },
    {
      "epoch": 3.6048741510187776,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 9023
    },
    {
      "epoch": 3.605273671594087,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 9024
    },
    {
      "epoch": 3.6056731921693967,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 9025
    },
    {
      "epoch": 3.606072712744706,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 9026
    },
    {
      "epoch": 3.606472233320016,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 9027
    },
    {
      "epoch": 3.6068717538953257,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 9028
    },
    {
      "epoch": 3.6072712744706354,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 9029
    },
    {
      "epoch": 3.607670795045945,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 9030
    },
    {
      "epoch": 3.6080703156212546,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 9031
    },
    {
      "epoch": 3.608469836196564,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 9032
    },
    {
      "epoch": 3.6088693567718737,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 9033
    },
    {
      "epoch": 3.6092688773471835,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 9034
    },
    {
      "epoch": 3.6096683979224933,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 9035
    },
    {
      "epoch": 3.6100679184978026,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 9036
    },
    {
      "epoch": 3.6104674390731124,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 9037
    },
    {
      "epoch": 3.6108669596484217,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 9038
    },
    {
      "epoch": 3.6112664802237315,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 9039
    },
    {
      "epoch": 3.6116660007990413,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 9040
    },
    {
      "epoch": 3.6120655213743507,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 9041
    },
    {
      "epoch": 3.6124650419496604,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 9042
    },
    {
      "epoch": 3.61286456252497,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 9043
    },
    {
      "epoch": 3.6132640831002796,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 9044
    },
    {
      "epoch": 3.6136636036755894,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 9045
    },
    {
      "epoch": 3.614063124250899,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 9046
    },
    {
      "epoch": 3.6144626448262085,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 9047
    },
    {
      "epoch": 3.6148621654015183,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 9048
    },
    {
      "epoch": 3.6152616859768276,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 9049
    },
    {
      "epoch": 3.6156612065521374,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 9050
    },
    {
      "epoch": 3.616060727127447,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 9051
    },
    {
      "epoch": 3.6164602477027565,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 9052
    },
    {
      "epoch": 3.6168597682780663,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 9053
    },
    {
      "epoch": 3.617259288853376,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 9054
    },
    {
      "epoch": 3.6176588094286855,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 9055
    },
    {
      "epoch": 3.6180583300039952,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 9056
    },
    {
      "epoch": 3.618457850579305,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 9057
    },
    {
      "epoch": 3.6188573711546144,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 9058
    },
    {
      "epoch": 3.619256891729924,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 9059
    },
    {
      "epoch": 3.6196564123052335,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 9060
    },
    {
      "epoch": 3.6200559328805433,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 9061
    },
    {
      "epoch": 3.620455453455853,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 9062
    },
    {
      "epoch": 3.620854974031163,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 9063
    },
    {
      "epoch": 3.621254494606472,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 9064
    },
    {
      "epoch": 3.621654015181782,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 9065
    },
    {
      "epoch": 3.6220535357570913,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 9066
    },
    {
      "epoch": 3.622453056332401,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 9067
    },
    {
      "epoch": 3.622852576907711,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 9068
    },
    {
      "epoch": 3.6232520974830202,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 9069
    },
    {
      "epoch": 3.62365161805833,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 9070
    },
    {
      "epoch": 3.6240511386336394,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 9071
    },
    {
      "epoch": 3.624450659208949,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 9072
    },
    {
      "epoch": 3.624850179784259,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 9073
    },
    {
      "epoch": 3.6252497003595687,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 9074
    },
    {
      "epoch": 3.625649220934878,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 9075
    },
    {
      "epoch": 3.626048741510188,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 9076
    },
    {
      "epoch": 3.626448262085497,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 9077
    },
    {
      "epoch": 3.626847782660807,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 9078
    },
    {
      "epoch": 3.6272473032361168,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 9079
    },
    {
      "epoch": 3.6276468238114266,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 9080
    },
    {
      "epoch": 3.628046344386736,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 9081
    },
    {
      "epoch": 3.6284458649620457,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 9082
    },
    {
      "epoch": 3.628845385537355,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 9083
    },
    {
      "epoch": 3.629244906112665,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 9084
    },
    {
      "epoch": 3.6296444266879746,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 9085
    },
    {
      "epoch": 3.630043947263284,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 9086
    },
    {
      "epoch": 3.6304434678385937,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 9087
    },
    {
      "epoch": 3.630842988413903,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 9088
    },
    {
      "epoch": 3.631242508989213,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 9089
    },
    {
      "epoch": 3.6316420295645226,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 9090
    },
    {
      "epoch": 3.6320415501398324,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 9091
    },
    {
      "epoch": 3.6324410707151418,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0644,
      "step": 9092
    },
    {
      "epoch": 3.6328405912904516,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 9093
    },
    {
      "epoch": 3.633240111865761,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 9094
    },
    {
      "epoch": 3.6336396324410707,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 9095
    },
    {
      "epoch": 3.6340391530163805,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 9096
    },
    {
      "epoch": 3.63443867359169,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 9097
    },
    {
      "epoch": 3.6348381941669996,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 9098
    },
    {
      "epoch": 3.6352377147423094,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 9099
    },
    {
      "epoch": 3.6356372353176187,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 9100
    },
    {
      "epoch": 3.6360367558929285,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 9101
    },
    {
      "epoch": 3.6364362764682383,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 9102
    },
    {
      "epoch": 3.6368357970435476,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 9103
    },
    {
      "epoch": 3.6372353176188574,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 9104
    },
    {
      "epoch": 3.6376348381941668,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 9105
    },
    {
      "epoch": 3.6380343587694766,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 9106
    },
    {
      "epoch": 3.6384338793447863,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 9107
    },
    {
      "epoch": 3.638833399920096,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 9108
    },
    {
      "epoch": 3.6392329204954055,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 9109
    },
    {
      "epoch": 3.6396324410707153,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 9110
    },
    {
      "epoch": 3.6400319616460246,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 9111
    },
    {
      "epoch": 3.6404314822213344,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 9112
    },
    {
      "epoch": 3.640831002796644,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 9113
    },
    {
      "epoch": 3.6412305233719535,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 9114
    },
    {
      "epoch": 3.6416300439472633,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 9115
    },
    {
      "epoch": 3.6420295645225726,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 9116
    },
    {
      "epoch": 3.6424290850978824,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 9117
    },
    {
      "epoch": 3.642828605673192,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 9118
    },
    {
      "epoch": 3.643228126248502,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 9119
    },
    {
      "epoch": 3.6436276468238114,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 9120
    },
    {
      "epoch": 3.644027167399121,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 9121
    },
    {
      "epoch": 3.6444266879744305,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 9122
    },
    {
      "epoch": 3.6448262085497403,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 9123
    },
    {
      "epoch": 3.64522572912505,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 9124
    },
    {
      "epoch": 3.64562524970036,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 9125
    },
    {
      "epoch": 3.646024770275669,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 9126
    },
    {
      "epoch": 3.646424290850979,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 9127
    },
    {
      "epoch": 3.6468238114262883,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 9128
    },
    {
      "epoch": 3.647223332001598,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 9129
    },
    {
      "epoch": 3.647622852576908,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 9130
    },
    {
      "epoch": 3.6480223731522172,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 9131
    },
    {
      "epoch": 3.648421893727527,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 9132
    },
    {
      "epoch": 3.6488214143028364,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 9133
    },
    {
      "epoch": 3.649220934878146,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 9134
    },
    {
      "epoch": 3.649620455453456,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 9135
    },
    {
      "epoch": 3.6500199760287657,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 9136
    },
    {
      "epoch": 3.650419496604075,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 9137
    },
    {
      "epoch": 3.650819017179385,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 9138
    },
    {
      "epoch": 3.651218537754694,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 9139
    },
    {
      "epoch": 3.651618058330004,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 9140
    },
    {
      "epoch": 3.6520175789053138,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 9141
    },
    {
      "epoch": 3.652417099480623,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 9142
    },
    {
      "epoch": 3.652816620055933,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 9143
    },
    {
      "epoch": 3.6532161406312427,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 9144
    },
    {
      "epoch": 3.653615661206552,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0689,
      "step": 9145
    },
    {
      "epoch": 3.654015181781862,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 9146
    },
    {
      "epoch": 3.6544147023571716,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 9147
    },
    {
      "epoch": 3.654814222932481,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 9148
    },
    {
      "epoch": 3.6552137435077907,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 9149
    },
    {
      "epoch": 3.6556132640831,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 9150
    },
    {
      "epoch": 3.65601278465841,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 9151
    },
    {
      "epoch": 3.6564123052337196,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 9152
    },
    {
      "epoch": 3.6568118258090294,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 9153
    },
    {
      "epoch": 3.6572113463843388,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 9154
    },
    {
      "epoch": 3.6576108669596485,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 9155
    },
    {
      "epoch": 3.658010387534958,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 9156
    },
    {
      "epoch": 3.6584099081102677,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 9157
    },
    {
      "epoch": 3.6588094286855775,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 9158
    },
    {
      "epoch": 3.659208949260887,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 9159
    },
    {
      "epoch": 3.6596084698361966,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 9160
    },
    {
      "epoch": 3.660007990411506,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 9161
    },
    {
      "epoch": 3.6604075109868157,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 9162
    },
    {
      "epoch": 3.6608070315621255,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 9163
    },
    {
      "epoch": 3.6612065521374353,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 9164
    },
    {
      "epoch": 3.6616060727127446,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 9165
    },
    {
      "epoch": 3.6620055932880544,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 9166
    },
    {
      "epoch": 3.6624051138633638,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 9167
    },
    {
      "epoch": 3.6628046344386735,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 9168
    },
    {
      "epoch": 3.6632041550139833,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 9169
    },
    {
      "epoch": 3.663603675589293,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 9170
    },
    {
      "epoch": 3.6640031961646025,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 9171
    },
    {
      "epoch": 3.6644027167399122,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 9172
    },
    {
      "epoch": 3.6648022373152216,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 9173
    },
    {
      "epoch": 3.6652017578905314,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 9174
    },
    {
      "epoch": 3.665601278465841,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 9175
    },
    {
      "epoch": 3.6660007990411505,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 9176
    },
    {
      "epoch": 3.6664003196164603,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 9177
    },
    {
      "epoch": 3.6667998401917696,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 9178
    },
    {
      "epoch": 3.6671993607670794,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 9179
    },
    {
      "epoch": 3.667598881342389,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 9180
    },
    {
      "epoch": 3.667998401917699,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 9181
    },
    {
      "epoch": 3.6683979224930083,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 9182
    },
    {
      "epoch": 3.668797443068318,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 9183
    },
    {
      "epoch": 3.6691969636436275,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 9184
    },
    {
      "epoch": 3.6695964842189372,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 9185
    },
    {
      "epoch": 3.669996004794247,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 9186
    },
    {
      "epoch": 3.6703955253695564,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 9187
    },
    {
      "epoch": 3.670795045944866,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 9188
    },
    {
      "epoch": 3.671194566520176,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 9189
    },
    {
      "epoch": 3.6715940870954853,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 9190
    },
    {
      "epoch": 3.671993607670795,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 9191
    },
    {
      "epoch": 3.672393128246105,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 9192
    },
    {
      "epoch": 3.672792648821414,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 9193
    },
    {
      "epoch": 3.673192169396724,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 9194
    },
    {
      "epoch": 3.6735916899720333,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 9195
    },
    {
      "epoch": 3.673991210547343,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 9196
    },
    {
      "epoch": 3.674390731122653,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 9197
    },
    {
      "epoch": 3.6747902516979627,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 9198
    },
    {
      "epoch": 3.675189772273272,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 9199
    },
    {
      "epoch": 3.675589292848582,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 9200
    },
    {
      "epoch": 3.675988813423891,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 9201
    },
    {
      "epoch": 3.676388333999201,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 9202
    },
    {
      "epoch": 3.6767878545745107,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 9203
    },
    {
      "epoch": 3.67718737514982,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 9204
    },
    {
      "epoch": 3.67758689572513,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 9205
    },
    {
      "epoch": 3.677986416300439,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 9206
    },
    {
      "epoch": 3.678385936875749,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 9207
    },
    {
      "epoch": 3.678785457451059,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 9208
    },
    {
      "epoch": 3.6791849780263686,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 9209
    },
    {
      "epoch": 3.679584498601678,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 9210
    },
    {
      "epoch": 3.6799840191769877,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 9211
    },
    {
      "epoch": 3.680383539752297,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 9212
    },
    {
      "epoch": 3.680783060327607,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 9213
    },
    {
      "epoch": 3.6811825809029166,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 9214
    },
    {
      "epoch": 3.6815821014782264,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 9215
    },
    {
      "epoch": 3.6819816220535357,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 9216
    },
    {
      "epoch": 3.6823811426288455,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 9217
    },
    {
      "epoch": 3.682780663204155,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 9218
    },
    {
      "epoch": 3.6831801837794647,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 9219
    },
    {
      "epoch": 3.6835797043547744,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 9220
    },
    {
      "epoch": 3.683979224930084,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 9221
    },
    {
      "epoch": 3.6843787455053936,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 9222
    },
    {
      "epoch": 3.684778266080703,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 9223
    },
    {
      "epoch": 3.6851777866560127,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 9224
    },
    {
      "epoch": 3.6855773072313225,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 9225
    },
    {
      "epoch": 3.6859768278066323,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 9226
    },
    {
      "epoch": 3.6863763483819416,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 9227
    },
    {
      "epoch": 3.6867758689572514,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 9228
    },
    {
      "epoch": 3.6871753895325607,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 9229
    },
    {
      "epoch": 3.6875749101078705,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 9230
    },
    {
      "epoch": 3.6879744306831803,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 9231
    },
    {
      "epoch": 3.6883739512584897,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 9232
    },
    {
      "epoch": 3.6887734718337994,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 9233
    },
    {
      "epoch": 3.6891729924091092,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 9234
    },
    {
      "epoch": 3.6895725129844186,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 9235
    },
    {
      "epoch": 3.6899720335597284,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 9236
    },
    {
      "epoch": 3.690371554135038,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 9237
    },
    {
      "epoch": 3.6907710747103475,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 9238
    },
    {
      "epoch": 3.6911705952856573,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 9239
    },
    {
      "epoch": 3.6915701158609666,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 9240
    },
    {
      "epoch": 3.6919696364362764,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 9241
    },
    {
      "epoch": 3.692369157011586,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 9242
    },
    {
      "epoch": 3.692768677586896,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 9243
    },
    {
      "epoch": 3.6931681981622053,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 9244
    },
    {
      "epoch": 3.693567718737515,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 9245
    },
    {
      "epoch": 3.6939672393128244,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 9246
    },
    {
      "epoch": 3.6943667598881342,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 9247
    },
    {
      "epoch": 3.694766280463444,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 9248
    },
    {
      "epoch": 3.6951658010387534,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 9249
    },
    {
      "epoch": 3.695565321614063,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0656,
      "step": 9250
    },
    {
      "epoch": 3.6959648421893725,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 9251
    },
    {
      "epoch": 3.6963643627646823,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 9252
    },
    {
      "epoch": 3.696763883339992,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 9253
    },
    {
      "epoch": 3.697163403915302,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 9254
    },
    {
      "epoch": 3.697562924490611,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 9255
    },
    {
      "epoch": 3.697962445065921,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 9256
    },
    {
      "epoch": 3.6983619656412303,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 9257
    },
    {
      "epoch": 3.69876148621654,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 9258
    },
    {
      "epoch": 3.69916100679185,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 9259
    },
    {
      "epoch": 3.6995605273671597,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 9260
    },
    {
      "epoch": 3.699960047942469,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 9261
    },
    {
      "epoch": 3.700359568517779,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 9262
    },
    {
      "epoch": 3.700759089093088,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 9263
    },
    {
      "epoch": 3.701158609668398,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 9264
    },
    {
      "epoch": 3.7015581302437077,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 9265
    },
    {
      "epoch": 3.701957650819017,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 9266
    },
    {
      "epoch": 3.702357171394327,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 9267
    },
    {
      "epoch": 3.702756691969636,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 9268
    },
    {
      "epoch": 3.703156212544946,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 9269
    },
    {
      "epoch": 3.7035557331202558,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 9270
    },
    {
      "epoch": 3.7039552536955656,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 9271
    },
    {
      "epoch": 3.704354774270875,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 9272
    },
    {
      "epoch": 3.7047542948461847,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 9273
    },
    {
      "epoch": 3.705153815421494,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 9274
    },
    {
      "epoch": 3.705553335996804,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 9275
    },
    {
      "epoch": 3.7059528565721136,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 9276
    },
    {
      "epoch": 3.706352377147423,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 9277
    },
    {
      "epoch": 3.7067518977227327,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 9278
    },
    {
      "epoch": 3.7071514182980425,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 9279
    },
    {
      "epoch": 3.707550938873352,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 9280
    },
    {
      "epoch": 3.7079504594486616,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 9281
    },
    {
      "epoch": 3.7083499800239714,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 9282
    },
    {
      "epoch": 3.7087495005992808,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 9283
    },
    {
      "epoch": 3.7091490211745906,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 9284
    },
    {
      "epoch": 3.7095485417499,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 9285
    },
    {
      "epoch": 3.7099480623252097,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 9286
    },
    {
      "epoch": 3.7103475829005195,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 9287
    },
    {
      "epoch": 3.7107471034758293,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 9288
    },
    {
      "epoch": 3.7111466240511386,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 9289
    },
    {
      "epoch": 3.7115461446264484,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 9290
    },
    {
      "epoch": 3.7119456652017577,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 9291
    },
    {
      "epoch": 3.7123451857770675,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 9292
    },
    {
      "epoch": 3.7127447063523773,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 9293
    },
    {
      "epoch": 3.7131442269276866,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 9294
    },
    {
      "epoch": 3.7135437475029964,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 9295
    },
    {
      "epoch": 3.7139432680783058,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 9296
    },
    {
      "epoch": 3.7143427886536156,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 9297
    },
    {
      "epoch": 3.7147423092289253,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 9298
    },
    {
      "epoch": 3.715141829804235,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 9299
    },
    {
      "epoch": 3.7155413503795445,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 9300
    },
    {
      "epoch": 3.7159408709548543,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 9301
    },
    {
      "epoch": 3.7163403915301636,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 9302
    },
    {
      "epoch": 3.7167399121054734,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 9303
    },
    {
      "epoch": 3.717139432680783,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 9304
    },
    {
      "epoch": 3.717538953256093,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 9305
    },
    {
      "epoch": 3.7179384738314023,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 9306
    },
    {
      "epoch": 3.718337994406712,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0661,
      "step": 9307
    },
    {
      "epoch": 3.7187375149820214,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 9308
    },
    {
      "epoch": 3.719137035557331,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 9309
    },
    {
      "epoch": 3.719536556132641,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 9310
    },
    {
      "epoch": 3.7199360767079503,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 9311
    },
    {
      "epoch": 3.72033559728326,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0651,
      "step": 9312
    },
    {
      "epoch": 3.7207351178585695,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 9313
    },
    {
      "epoch": 3.7211346384338793,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 9314
    },
    {
      "epoch": 3.721534159009189,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 9315
    },
    {
      "epoch": 3.721933679584499,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 9316
    },
    {
      "epoch": 3.722333200159808,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 9317
    },
    {
      "epoch": 3.722732720735118,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 9318
    },
    {
      "epoch": 3.7231322413104273,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 9319
    },
    {
      "epoch": 3.723531761885737,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 9320
    },
    {
      "epoch": 3.723931282461047,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 9321
    },
    {
      "epoch": 3.724330803036356,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 9322
    },
    {
      "epoch": 3.724730323611666,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 9323
    },
    {
      "epoch": 3.725129844186976,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 9324
    },
    {
      "epoch": 3.725529364762285,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 9325
    },
    {
      "epoch": 3.725928885337595,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 9326
    },
    {
      "epoch": 3.7263284059129047,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 9327
    },
    {
      "epoch": 3.726727926488214,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 9328
    },
    {
      "epoch": 3.727127447063524,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 9329
    },
    {
      "epoch": 3.727526967638833,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 9330
    },
    {
      "epoch": 3.727926488214143,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 9331
    },
    {
      "epoch": 3.7283260087894528,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 9332
    },
    {
      "epoch": 3.7287255293647625,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 9333
    },
    {
      "epoch": 3.729125049940072,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 9334
    },
    {
      "epoch": 3.7295245705153817,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 9335
    },
    {
      "epoch": 3.729924091090691,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 9336
    },
    {
      "epoch": 3.730323611666001,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 9337
    },
    {
      "epoch": 3.7307231322413106,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 9338
    },
    {
      "epoch": 3.73112265281662,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 9339
    },
    {
      "epoch": 3.7315221733919297,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 9340
    },
    {
      "epoch": 3.731921693967239,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 9341
    },
    {
      "epoch": 3.732321214542549,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 9342
    },
    {
      "epoch": 3.7327207351178586,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 9343
    },
    {
      "epoch": 3.7331202556931684,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 9344
    },
    {
      "epoch": 3.7335197762684778,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 9345
    },
    {
      "epoch": 3.7339192968437875,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 9346
    },
    {
      "epoch": 3.734318817419097,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 9347
    },
    {
      "epoch": 3.7347183379944067,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 9348
    },
    {
      "epoch": 3.7351178585697165,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 9349
    },
    {
      "epoch": 3.7355173791450262,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 9350
    },
    {
      "epoch": 3.7359168997203356,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 9351
    },
    {
      "epoch": 3.7363164202956454,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 9352
    },
    {
      "epoch": 3.7367159408709547,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 9353
    },
    {
      "epoch": 3.7371154614462645,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 9354
    },
    {
      "epoch": 3.7375149820215743,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 9355
    },
    {
      "epoch": 3.7379145025968836,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 9356
    },
    {
      "epoch": 3.7383140231721934,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 9357
    },
    {
      "epoch": 3.7387135437475028,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 9358
    },
    {
      "epoch": 3.7391130643228125,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 9359
    },
    {
      "epoch": 3.7395125848981223,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 9360
    },
    {
      "epoch": 3.739912105473432,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 9361
    },
    {
      "epoch": 3.7403116260487415,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 9362
    },
    {
      "epoch": 3.7407111466240512,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0673,
      "step": 9363
    },
    {
      "epoch": 3.7411106671993606,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 9364
    },
    {
      "epoch": 3.7415101877746704,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 9365
    },
    {
      "epoch": 3.74190970834998,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 9366
    },
    {
      "epoch": 3.7423092289252895,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 9367
    },
    {
      "epoch": 3.7427087495005993,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 9368
    },
    {
      "epoch": 3.743108270075909,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 9369
    },
    {
      "epoch": 3.7435077906512184,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 9370
    },
    {
      "epoch": 3.743907311226528,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 9371
    },
    {
      "epoch": 3.744306831801838,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 9372
    },
    {
      "epoch": 3.7447063523771473,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 9373
    },
    {
      "epoch": 3.745105872952457,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 9374
    },
    {
      "epoch": 3.7455053935277665,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 9375
    },
    {
      "epoch": 3.7459049141030762,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 9376
    },
    {
      "epoch": 3.746304434678386,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 9377
    },
    {
      "epoch": 3.746703955253696,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 9378
    },
    {
      "epoch": 3.747103475829005,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 9379
    },
    {
      "epoch": 3.747502996404315,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 9380
    },
    {
      "epoch": 3.7479025169796243,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 9381
    },
    {
      "epoch": 3.748302037554934,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 9382
    },
    {
      "epoch": 3.748701558130244,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 9383
    },
    {
      "epoch": 3.749101078705553,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0598,
      "step": 9384
    },
    {
      "epoch": 3.749500599280863,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 9385
    },
    {
      "epoch": 3.7499001198561723,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 9386
    },
    {
      "epoch": 3.750299640431482,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 9387
    },
    {
      "epoch": 3.750699161006792,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 9388
    },
    {
      "epoch": 3.7510986815821017,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 9389
    },
    {
      "epoch": 3.751498202157411,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 9390
    },
    {
      "epoch": 3.751897722732721,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 9391
    },
    {
      "epoch": 3.75229724330803,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 9392
    },
    {
      "epoch": 3.75269676388334,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 9393
    },
    {
      "epoch": 3.7530962844586497,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 9394
    },
    {
      "epoch": 3.7534958050339595,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 9395
    },
    {
      "epoch": 3.753895325609269,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 9396
    },
    {
      "epoch": 3.7542948461845786,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 9397
    },
    {
      "epoch": 3.754694366759888,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 9398
    },
    {
      "epoch": 3.7550938873351978,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 9399
    },
    {
      "epoch": 3.7554934079105076,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 9400
    },
    {
      "epoch": 3.755892928485817,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 9401
    },
    {
      "epoch": 3.7562924490611267,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 9402
    },
    {
      "epoch": 3.756691969636436,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 9403
    },
    {
      "epoch": 3.757091490211746,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 9404
    },
    {
      "epoch": 3.7574910107870556,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 9405
    },
    {
      "epoch": 3.7578905313623654,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 9406
    },
    {
      "epoch": 3.7582900519376747,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 9407
    },
    {
      "epoch": 3.7586895725129845,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 9408
    },
    {
      "epoch": 3.759089093088294,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 9409
    },
    {
      "epoch": 3.7594886136636037,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 9410
    },
    {
      "epoch": 3.7598881342389134,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 9411
    },
    {
      "epoch": 3.760287654814223,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 9412
    },
    {
      "epoch": 3.7606871753895326,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 9413
    },
    {
      "epoch": 3.7610866959648424,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 9414
    },
    {
      "epoch": 3.7614862165401517,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 9415
    },
    {
      "epoch": 3.7618857371154615,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 9416
    },
    {
      "epoch": 3.7622852576907713,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 9417
    },
    {
      "epoch": 3.7626847782660806,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 9418
    },
    {
      "epoch": 3.7630842988413904,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 9419
    },
    {
      "epoch": 3.7634838194166997,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 9420
    },
    {
      "epoch": 3.7638833399920095,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 9421
    },
    {
      "epoch": 3.7642828605673193,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 9422
    },
    {
      "epoch": 3.764682381142629,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 9423
    },
    {
      "epoch": 3.7650819017179384,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 9424
    },
    {
      "epoch": 3.7654814222932482,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 9425
    },
    {
      "epoch": 3.7658809428685576,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 9426
    },
    {
      "epoch": 3.7662804634438674,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 9427
    },
    {
      "epoch": 3.766679984019177,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 9428
    },
    {
      "epoch": 3.7670795045944865,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 9429
    },
    {
      "epoch": 3.7674790251697963,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 9430
    },
    {
      "epoch": 3.7678785457451056,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 9431
    },
    {
      "epoch": 3.7682780663204154,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 9432
    },
    {
      "epoch": 3.768677586895725,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 9433
    },
    {
      "epoch": 3.769077107471035,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 9434
    },
    {
      "epoch": 3.7694766280463443,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 9435
    },
    {
      "epoch": 3.769876148621654,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 9436
    },
    {
      "epoch": 3.7702756691969634,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 9437
    },
    {
      "epoch": 3.7706751897722732,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 9438
    },
    {
      "epoch": 3.771074710347583,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 9439
    },
    {
      "epoch": 3.771474230922893,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 9440
    },
    {
      "epoch": 3.771873751498202,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 9441
    },
    {
      "epoch": 3.772273272073512,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 9442
    },
    {
      "epoch": 3.7726727926488213,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 9443
    },
    {
      "epoch": 3.773072313224131,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 9444
    },
    {
      "epoch": 3.773471833799441,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 9445
    },
    {
      "epoch": 3.77387135437475,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 9446
    },
    {
      "epoch": 3.77427087495006,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 9447
    },
    {
      "epoch": 3.7746703955253693,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 9448
    },
    {
      "epoch": 3.775069916100679,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 9449
    },
    {
      "epoch": 3.775469436675989,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 9450
    },
    {
      "epoch": 3.7758689572512987,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 9451
    },
    {
      "epoch": 3.776268477826608,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 9452
    },
    {
      "epoch": 3.776667998401918,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 9453
    },
    {
      "epoch": 3.777067518977227,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 9454
    },
    {
      "epoch": 3.777467039552537,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 9455
    },
    {
      "epoch": 3.7778665601278467,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 9456
    },
    {
      "epoch": 3.778266080703156,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 9457
    },
    {
      "epoch": 3.778665601278466,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 9458
    },
    {
      "epoch": 3.7790651218537756,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 9459
    },
    {
      "epoch": 3.779464642429085,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 9460
    },
    {
      "epoch": 3.7798641630043948,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 9461
    },
    {
      "epoch": 3.7802636835797045,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 9462
    },
    {
      "epoch": 3.780663204155014,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 9463
    },
    {
      "epoch": 3.7810627247303237,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 9464
    },
    {
      "epoch": 3.781462245305633,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 9465
    },
    {
      "epoch": 3.781861765880943,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 9466
    },
    {
      "epoch": 3.7822612864562526,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 9467
    },
    {
      "epoch": 3.7826608070315624,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 9468
    },
    {
      "epoch": 3.7830603276068717,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 9469
    },
    {
      "epoch": 3.7834598481821815,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 9470
    },
    {
      "epoch": 3.783859368757491,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 9471
    },
    {
      "epoch": 3.7842588893328006,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 9472
    },
    {
      "epoch": 3.7846584099081104,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 9473
    },
    {
      "epoch": 3.7850579304834198,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 9474
    },
    {
      "epoch": 3.7854574510587295,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 9475
    },
    {
      "epoch": 3.785856971634039,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 9476
    },
    {
      "epoch": 3.7862564922093487,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 9477
    },
    {
      "epoch": 3.7866560127846585,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 9478
    },
    {
      "epoch": 3.7870555333599683,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 9479
    },
    {
      "epoch": 3.7874550539352776,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 9480
    },
    {
      "epoch": 3.7878545745105874,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 9481
    },
    {
      "epoch": 3.7882540950858967,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 9482
    },
    {
      "epoch": 3.7886536156612065,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 9483
    },
    {
      "epoch": 3.7890531362365163,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 9484
    },
    {
      "epoch": 3.789452656811826,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 9485
    },
    {
      "epoch": 3.7898521773871354,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 9486
    },
    {
      "epoch": 3.790251697962445,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 9487
    },
    {
      "epoch": 3.7906512185377546,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 9488
    },
    {
      "epoch": 3.7910507391130643,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 9489
    },
    {
      "epoch": 3.791450259688374,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 9490
    },
    {
      "epoch": 3.7918497802636835,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 9491
    },
    {
      "epoch": 3.7922493008389933,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 9492
    },
    {
      "epoch": 3.7926488214143026,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 9493
    },
    {
      "epoch": 3.7930483419896124,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 9494
    },
    {
      "epoch": 3.793447862564922,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 9495
    },
    {
      "epoch": 3.793847383140232,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 9496
    },
    {
      "epoch": 3.7942469037155413,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 9497
    },
    {
      "epoch": 3.794646424290851,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 9498
    },
    {
      "epoch": 3.7950459448661604,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 9499
    },
    {
      "epoch": 3.79544546544147,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 9500
    },
    {
      "epoch": 3.79584498601678,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 9501
    },
    {
      "epoch": 3.7962445065920893,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 9502
    },
    {
      "epoch": 3.796644027167399,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 9503
    },
    {
      "epoch": 3.797043547742709,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 9504
    },
    {
      "epoch": 3.7974430683180183,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 9505
    },
    {
      "epoch": 3.797842588893328,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 9506
    },
    {
      "epoch": 3.798242109468638,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 9507
    },
    {
      "epoch": 3.798641630043947,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 9508
    },
    {
      "epoch": 3.799041150619257,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 9509
    },
    {
      "epoch": 3.7994406711945663,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 9510
    },
    {
      "epoch": 3.799840191769876,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 9511
    },
    {
      "epoch": 3.800239712345186,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 9512
    },
    {
      "epoch": 3.8006392329204957,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 9513
    },
    {
      "epoch": 3.801038753495805,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 9514
    },
    {
      "epoch": 3.801438274071115,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 9515
    },
    {
      "epoch": 3.801837794646424,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 9516
    },
    {
      "epoch": 3.802237315221734,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 9517
    },
    {
      "epoch": 3.8026368357970437,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 9518
    },
    {
      "epoch": 3.803036356372353,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 9519
    },
    {
      "epoch": 3.803435876947663,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 9520
    },
    {
      "epoch": 3.803835397522972,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 9521
    },
    {
      "epoch": 3.804234918098282,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 9522
    },
    {
      "epoch": 3.8046344386735917,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 9523
    },
    {
      "epoch": 3.8050339592489015,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 9524
    },
    {
      "epoch": 3.805433479824211,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 9525
    },
    {
      "epoch": 3.8058330003995207,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 9526
    },
    {
      "epoch": 3.80623252097483,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 9527
    },
    {
      "epoch": 3.80663204155014,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 9528
    },
    {
      "epoch": 3.8070315621254496,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 9529
    },
    {
      "epoch": 3.8074310827007594,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 9530
    },
    {
      "epoch": 3.8078306032760687,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 9531
    },
    {
      "epoch": 3.8082301238513785,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 9532
    },
    {
      "epoch": 3.808629644426688,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 9533
    },
    {
      "epoch": 3.8090291650019976,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 9534
    },
    {
      "epoch": 3.8094286855773074,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 9535
    },
    {
      "epoch": 3.8098282061526167,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 9536
    },
    {
      "epoch": 3.8102277267279265,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 9537
    },
    {
      "epoch": 3.810627247303236,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 9538
    },
    {
      "epoch": 3.8110267678785457,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 9539
    },
    {
      "epoch": 3.8114262884538554,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 9540
    },
    {
      "epoch": 3.8118258090291652,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 9541
    },
    {
      "epoch": 3.8122253296044746,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 9542
    },
    {
      "epoch": 3.8126248501797844,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 9543
    },
    {
      "epoch": 3.8130243707550937,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 9544
    },
    {
      "epoch": 3.8134238913304035,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 9545
    },
    {
      "epoch": 3.8138234119057133,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 9546
    },
    {
      "epoch": 3.8142229324810226,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 9547
    },
    {
      "epoch": 3.8146224530563324,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 9548
    },
    {
      "epoch": 3.815021973631642,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 9549
    },
    {
      "epoch": 3.8154214942069515,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 9550
    },
    {
      "epoch": 3.8158210147822613,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 9551
    },
    {
      "epoch": 3.816220535357571,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 9552
    },
    {
      "epoch": 3.8166200559328805,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 9553
    },
    {
      "epoch": 3.8170195765081902,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 9554
    },
    {
      "epoch": 3.8174190970834996,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 9555
    },
    {
      "epoch": 3.8178186176588094,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 9556
    },
    {
      "epoch": 3.818218138234119,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 9557
    },
    {
      "epoch": 3.818617658809429,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 9558
    },
    {
      "epoch": 3.8190171793847383,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 9559
    },
    {
      "epoch": 3.819416699960048,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 9560
    },
    {
      "epoch": 3.8198162205353574,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 9561
    },
    {
      "epoch": 3.820215741110667,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 9562
    },
    {
      "epoch": 3.820615261685977,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 9563
    },
    {
      "epoch": 3.8210147822612863,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 9564
    },
    {
      "epoch": 3.821414302836596,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 9565
    },
    {
      "epoch": 3.8218138234119055,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 9566
    },
    {
      "epoch": 3.8222133439872152,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 9567
    },
    {
      "epoch": 3.822612864562525,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 9568
    },
    {
      "epoch": 3.823012385137835,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 9569
    },
    {
      "epoch": 3.823411905713144,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 9570
    },
    {
      "epoch": 3.823811426288454,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 9571
    },
    {
      "epoch": 3.8242109468637633,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 9572
    },
    {
      "epoch": 3.824610467439073,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 9573
    },
    {
      "epoch": 3.825009988014383,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 9574
    },
    {
      "epoch": 3.8254095085896926,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 9575
    },
    {
      "epoch": 3.825809029165002,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 9576
    },
    {
      "epoch": 3.8262085497403118,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 9577
    },
    {
      "epoch": 3.826608070315621,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 9578
    },
    {
      "epoch": 3.827007590890931,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0653,
      "step": 9579
    },
    {
      "epoch": 3.8274071114662407,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 9580
    },
    {
      "epoch": 3.82780663204155,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 9581
    },
    {
      "epoch": 3.82820615261686,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 9582
    },
    {
      "epoch": 3.828605673192169,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 9583
    },
    {
      "epoch": 3.829005193767479,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 9584
    },
    {
      "epoch": 3.8294047143427887,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 9585
    },
    {
      "epoch": 3.8298042349180985,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 9586
    },
    {
      "epoch": 3.830203755493408,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 9587
    },
    {
      "epoch": 3.8306032760687176,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 9588
    },
    {
      "epoch": 3.831002796644027,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 9589
    },
    {
      "epoch": 3.8314023172193368,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 9590
    },
    {
      "epoch": 3.8318018377946466,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 9591
    },
    {
      "epoch": 3.832201358369956,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 9592
    },
    {
      "epoch": 3.8326008789452657,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 9593
    },
    {
      "epoch": 3.8330003995205755,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 9594
    },
    {
      "epoch": 3.833399920095885,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 9595
    },
    {
      "epoch": 3.8337994406711946,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 9596
    },
    {
      "epoch": 3.8341989612465044,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 9597
    },
    {
      "epoch": 3.8345984818218137,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 9598
    },
    {
      "epoch": 3.8349980023971235,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 9599
    },
    {
      "epoch": 3.835397522972433,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 9600
    },
    {
      "epoch": 3.8357970435477426,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 9601
    },
    {
      "epoch": 3.8361965641230524,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 9602
    },
    {
      "epoch": 3.836596084698362,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 9603
    },
    {
      "epoch": 3.8369956052736716,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 9604
    },
    {
      "epoch": 3.8373951258489813,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 9605
    },
    {
      "epoch": 3.8377946464242907,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 9606
    },
    {
      "epoch": 3.8381941669996005,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 9607
    },
    {
      "epoch": 3.8385936875749103,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 9608
    },
    {
      "epoch": 3.8389932081502196,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 9609
    },
    {
      "epoch": 3.8393927287255294,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 9610
    },
    {
      "epoch": 3.8397922493008387,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 9611
    },
    {
      "epoch": 3.8401917698761485,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 9612
    },
    {
      "epoch": 3.8405912904514583,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 9613
    },
    {
      "epoch": 3.840990811026768,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 9614
    },
    {
      "epoch": 3.8413903316020774,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 9615
    },
    {
      "epoch": 3.841789852177387,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 9616
    },
    {
      "epoch": 3.8421893727526966,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 9617
    },
    {
      "epoch": 3.8425888933280063,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 9618
    },
    {
      "epoch": 3.842988413903316,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 9619
    },
    {
      "epoch": 3.843387934478626,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 9620
    },
    {
      "epoch": 3.8437874550539353,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 9621
    },
    {
      "epoch": 3.844186975629245,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 9622
    },
    {
      "epoch": 3.8445864962045544,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 9623
    },
    {
      "epoch": 3.844986016779864,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 9624
    },
    {
      "epoch": 3.845385537355174,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 9625
    },
    {
      "epoch": 3.8457850579304833,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 9626
    },
    {
      "epoch": 3.846184578505793,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 9627
    },
    {
      "epoch": 3.8465840990811024,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 9628
    },
    {
      "epoch": 3.8469836196564122,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 9629
    },
    {
      "epoch": 3.847383140231722,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 9630
    },
    {
      "epoch": 3.847782660807032,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 9631
    },
    {
      "epoch": 3.848182181382341,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 9632
    },
    {
      "epoch": 3.848581701957651,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 9633
    },
    {
      "epoch": 3.8489812225329603,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 9634
    },
    {
      "epoch": 3.84938074310827,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 9635
    },
    {
      "epoch": 3.84978026368358,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 9636
    },
    {
      "epoch": 3.850179784258889,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 9637
    },
    {
      "epoch": 3.850579304834199,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 9638
    },
    {
      "epoch": 3.8509788254095088,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 9639
    },
    {
      "epoch": 3.851378345984818,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 9640
    },
    {
      "epoch": 3.851777866560128,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 9641
    },
    {
      "epoch": 3.8521773871354377,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 9642
    },
    {
      "epoch": 3.852576907710747,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 9643
    },
    {
      "epoch": 3.852976428286057,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 9644
    },
    {
      "epoch": 3.853375948861366,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 9645
    },
    {
      "epoch": 3.853775469436676,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 9646
    },
    {
      "epoch": 3.8541749900119857,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 9647
    },
    {
      "epoch": 3.8545745105872955,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 9648
    },
    {
      "epoch": 3.854974031162605,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 9649
    },
    {
      "epoch": 3.8553735517379146,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 9650
    },
    {
      "epoch": 3.855773072313224,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 9651
    },
    {
      "epoch": 3.8561725928885338,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 9652
    },
    {
      "epoch": 3.8565721134638435,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 9653
    },
    {
      "epoch": 3.856971634039153,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 9654
    },
    {
      "epoch": 3.8573711546144627,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 9655
    },
    {
      "epoch": 3.857770675189772,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 9656
    },
    {
      "epoch": 3.858170195765082,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 9657
    },
    {
      "epoch": 3.8585697163403916,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 9658
    },
    {
      "epoch": 3.8589692369157014,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 9659
    },
    {
      "epoch": 3.8593687574910107,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 9660
    },
    {
      "epoch": 3.8597682780663205,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 9661
    },
    {
      "epoch": 3.86016779864163,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 9662
    },
    {
      "epoch": 3.8605673192169396,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 9663
    },
    {
      "epoch": 3.8609668397922494,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 9664
    },
    {
      "epoch": 3.861366360367559,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 9665
    },
    {
      "epoch": 3.8617658809428685,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 9666
    },
    {
      "epoch": 3.8621654015181783,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 9667
    },
    {
      "epoch": 3.8625649220934877,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 9668
    },
    {
      "epoch": 3.8629644426687975,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 9669
    },
    {
      "epoch": 3.8633639632441072,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 9670
    },
    {
      "epoch": 3.8637634838194166,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 9671
    },
    {
      "epoch": 3.8641630043947264,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 9672
    },
    {
      "epoch": 3.8645625249700357,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 9673
    },
    {
      "epoch": 3.8649620455453455,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 9674
    },
    {
      "epoch": 3.8653615661206553,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 9675
    },
    {
      "epoch": 3.865761086695965,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 9676
    },
    {
      "epoch": 3.8661606072712744,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 9677
    },
    {
      "epoch": 3.866560127846584,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 9678
    },
    {
      "epoch": 3.8669596484218935,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 9679
    },
    {
      "epoch": 3.8673591689972033,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 9680
    },
    {
      "epoch": 3.867758689572513,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 9681
    },
    {
      "epoch": 3.8681582101478225,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 9682
    },
    {
      "epoch": 3.8685577307231322,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 9683
    },
    {
      "epoch": 3.868957251298442,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 9684
    },
    {
      "epoch": 3.8693567718737514,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 9685
    },
    {
      "epoch": 3.869756292449061,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 9686
    },
    {
      "epoch": 3.870155813024371,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 9687
    },
    {
      "epoch": 3.8705553335996803,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 9688
    },
    {
      "epoch": 3.87095485417499,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 9689
    },
    {
      "epoch": 3.8713543747502994,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 9690
    },
    {
      "epoch": 3.871753895325609,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 9691
    },
    {
      "epoch": 3.872153415900919,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 9692
    },
    {
      "epoch": 3.872552936476229,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 9693
    },
    {
      "epoch": 3.872952457051538,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 9694
    },
    {
      "epoch": 3.873351977626848,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 9695
    },
    {
      "epoch": 3.8737514982021573,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 9696
    },
    {
      "epoch": 3.874151018777467,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 9697
    },
    {
      "epoch": 3.874550539352777,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 9698
    },
    {
      "epoch": 3.874950059928086,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 9699
    },
    {
      "epoch": 3.875349580503396,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 9700
    },
    {
      "epoch": 3.8757491010787053,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 9701
    },
    {
      "epoch": 3.876148621654015,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 9702
    },
    {
      "epoch": 3.876548142229325,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 9703
    },
    {
      "epoch": 3.8769476628046347,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 9704
    },
    {
      "epoch": 3.877347183379944,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 9705
    },
    {
      "epoch": 3.877746703955254,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 9706
    },
    {
      "epoch": 3.878146224530563,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 9707
    },
    {
      "epoch": 3.878545745105873,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 9708
    },
    {
      "epoch": 3.8789452656811827,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 9709
    },
    {
      "epoch": 3.8793447862564925,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 9710
    },
    {
      "epoch": 3.879744306831802,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 9711
    },
    {
      "epoch": 3.8801438274071116,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 9712
    },
    {
      "epoch": 3.880543347982421,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 9713
    },
    {
      "epoch": 3.8809428685577307,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 9714
    },
    {
      "epoch": 3.8813423891330405,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 9715
    },
    {
      "epoch": 3.88174190970835,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 9716
    },
    {
      "epoch": 3.8821414302836597,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 9717
    },
    {
      "epoch": 3.882540950858969,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 9718
    },
    {
      "epoch": 3.882940471434279,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 9719
    },
    {
      "epoch": 3.8833399920095886,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 9720
    },
    {
      "epoch": 3.8837395125848984,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 9721
    },
    {
      "epoch": 3.8841390331602077,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 9722
    },
    {
      "epoch": 3.8845385537355175,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 9723
    },
    {
      "epoch": 3.884938074310827,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 9724
    },
    {
      "epoch": 3.8853375948861366,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 9725
    },
    {
      "epoch": 3.8857371154614464,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 9726
    },
    {
      "epoch": 3.8861366360367557,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 9727
    },
    {
      "epoch": 3.8865361566120655,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 9728
    },
    {
      "epoch": 3.8869356771873753,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 9729
    },
    {
      "epoch": 3.8873351977626847,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 9730
    },
    {
      "epoch": 3.8877347183379944,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 9731
    },
    {
      "epoch": 3.8881342389133042,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 9732
    },
    {
      "epoch": 3.8885337594886136,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 9733
    },
    {
      "epoch": 3.8889332800639234,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 9734
    },
    {
      "epoch": 3.8893328006392327,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 9735
    },
    {
      "epoch": 3.8897323212145425,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 9736
    },
    {
      "epoch": 3.8901318417898523,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 9737
    },
    {
      "epoch": 3.890531362365162,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 9738
    },
    {
      "epoch": 3.8909308829404714,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 9739
    },
    {
      "epoch": 3.891330403515781,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 9740
    },
    {
      "epoch": 3.8917299240910905,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 9741
    },
    {
      "epoch": 3.8921294446664003,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 9742
    },
    {
      "epoch": 3.89252896524171,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 9743
    },
    {
      "epoch": 3.8929284858170194,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 9744
    },
    {
      "epoch": 3.8933280063923292,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 9745
    },
    {
      "epoch": 3.8937275269676386,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 9746
    },
    {
      "epoch": 3.8941270475429484,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 9747
    },
    {
      "epoch": 3.894526568118258,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 9748
    },
    {
      "epoch": 3.894926088693568,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 9749
    },
    {
      "epoch": 3.8953256092688773,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 9750
    },
    {
      "epoch": 3.895725129844187,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 9751
    },
    {
      "epoch": 3.8961246504194964,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 9752
    },
    {
      "epoch": 3.896524170994806,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 9753
    },
    {
      "epoch": 3.896923691570116,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 9754
    },
    {
      "epoch": 3.8973232121454258,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 9755
    },
    {
      "epoch": 3.897722732720735,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 9756
    },
    {
      "epoch": 3.898122253296045,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 9757
    },
    {
      "epoch": 3.8985217738713542,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 9758
    },
    {
      "epoch": 3.898921294446664,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 9759
    },
    {
      "epoch": 3.899320815021974,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 9760
    },
    {
      "epoch": 3.899720335597283,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 9761
    },
    {
      "epoch": 3.900119856172593,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 9762
    },
    {
      "epoch": 3.9005193767479023,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 9763
    },
    {
      "epoch": 3.900918897323212,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 9764
    },
    {
      "epoch": 3.901318417898522,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 9765
    },
    {
      "epoch": 3.9017179384738316,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 9766
    },
    {
      "epoch": 3.902117459049141,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 9767
    },
    {
      "epoch": 3.9025169796244508,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 9768
    },
    {
      "epoch": 3.90291650019976,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 9769
    },
    {
      "epoch": 3.90331602077507,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 9770
    },
    {
      "epoch": 3.9037155413503797,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 9771
    },
    {
      "epoch": 3.904115061925689,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 9772
    },
    {
      "epoch": 3.904514582500999,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 9773
    },
    {
      "epoch": 3.9049141030763086,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 9774
    },
    {
      "epoch": 3.905313623651618,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 9775
    },
    {
      "epoch": 3.9057131442269277,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 9776
    },
    {
      "epoch": 3.9061126648022375,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 9777
    },
    {
      "epoch": 3.906512185377547,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 9778
    },
    {
      "epoch": 3.9069117059528566,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 9779
    },
    {
      "epoch": 3.907311226528166,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 9780
    },
    {
      "epoch": 3.9077107471034758,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 9781
    },
    {
      "epoch": 3.9081102676787856,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 9782
    },
    {
      "epoch": 3.9085097882540953,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 9783
    },
    {
      "epoch": 3.9089093088294047,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 9784
    },
    {
      "epoch": 3.9093088294047145,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 9785
    },
    {
      "epoch": 3.909708349980024,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 9786
    },
    {
      "epoch": 3.9101078705553336,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 9787
    },
    {
      "epoch": 3.9105073911306434,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 9788
    },
    {
      "epoch": 3.9109069117059527,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 9789
    },
    {
      "epoch": 3.9113064322812625,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 9790
    },
    {
      "epoch": 3.911705952856572,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 9791
    },
    {
      "epoch": 3.9121054734318816,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 9792
    },
    {
      "epoch": 3.9125049940071914,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 9793
    },
    {
      "epoch": 3.912904514582501,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 9794
    },
    {
      "epoch": 3.9133040351578106,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 9795
    },
    {
      "epoch": 3.9137035557331203,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 9796
    },
    {
      "epoch": 3.9141030763084297,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 9797
    },
    {
      "epoch": 3.9145025968837395,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 9798
    },
    {
      "epoch": 3.9149021174590493,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 9799
    },
    {
      "epoch": 3.915301638034359,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 9800
    },
    {
      "epoch": 3.9157011586096684,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 9801
    },
    {
      "epoch": 3.916100679184978,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 9802
    },
    {
      "epoch": 3.9165001997602875,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 9803
    },
    {
      "epoch": 3.9168997203355973,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 9804
    },
    {
      "epoch": 3.917299240910907,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 9805
    },
    {
      "epoch": 3.9176987614862164,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 9806
    },
    {
      "epoch": 3.918098282061526,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 9807
    },
    {
      "epoch": 3.9184978026368356,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0692,
      "step": 9808
    },
    {
      "epoch": 3.9188973232121453,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0664,
      "step": 9809
    },
    {
      "epoch": 3.919296843787455,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 9810
    },
    {
      "epoch": 3.919696364362765,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 9811
    },
    {
      "epoch": 3.9200958849380743,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 9812
    },
    {
      "epoch": 3.920495405513384,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 9813
    },
    {
      "epoch": 3.9208949260886934,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 9814
    },
    {
      "epoch": 3.921294446664003,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 9815
    },
    {
      "epoch": 3.921693967239313,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 9816
    },
    {
      "epoch": 3.9220934878146223,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 9817
    },
    {
      "epoch": 3.922493008389932,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 9818
    },
    {
      "epoch": 3.922892528965242,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0654,
      "step": 9819
    },
    {
      "epoch": 3.923292049540551,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 9820
    },
    {
      "epoch": 3.923691570115861,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 9821
    },
    {
      "epoch": 3.924091090691171,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 9822
    },
    {
      "epoch": 3.92449061126648,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 9823
    },
    {
      "epoch": 3.92489013184179,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 9824
    },
    {
      "epoch": 3.9252896524170993,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 9825
    },
    {
      "epoch": 3.925689172992409,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 9826
    },
    {
      "epoch": 3.926088693567719,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 9827
    },
    {
      "epoch": 3.9264882141430286,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 9828
    },
    {
      "epoch": 3.926887734718338,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 9829
    },
    {
      "epoch": 3.9272872552936477,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 9830
    },
    {
      "epoch": 3.927686775868957,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 9831
    },
    {
      "epoch": 3.928086296444267,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 9832
    },
    {
      "epoch": 3.9284858170195767,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 9833
    },
    {
      "epoch": 3.928885337594886,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 9834
    },
    {
      "epoch": 3.929284858170196,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 9835
    },
    {
      "epoch": 3.929684378745505,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 9836
    },
    {
      "epoch": 3.930083899320815,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 9837
    },
    {
      "epoch": 3.9304834198961247,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 9838
    },
    {
      "epoch": 3.9308829404714345,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 9839
    },
    {
      "epoch": 3.931282461046744,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 9840
    },
    {
      "epoch": 3.9316819816220536,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 9841
    },
    {
      "epoch": 3.932081502197363,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 9842
    },
    {
      "epoch": 3.9324810227726728,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 9843
    },
    {
      "epoch": 3.9328805433479825,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 9844
    },
    {
      "epoch": 3.9332800639232923,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 9845
    },
    {
      "epoch": 3.9336795844986017,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 9846
    },
    {
      "epoch": 3.9340791050739115,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 9847
    },
    {
      "epoch": 3.934478625649221,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 9848
    },
    {
      "epoch": 3.9348781462245306,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 9849
    },
    {
      "epoch": 3.9352776667998404,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 9850
    },
    {
      "epoch": 3.9356771873751497,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 9851
    },
    {
      "epoch": 3.9360767079504595,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 9852
    },
    {
      "epoch": 3.936476228525769,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 9853
    },
    {
      "epoch": 3.9368757491010786,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 9854
    },
    {
      "epoch": 3.9372752696763884,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 9855
    },
    {
      "epoch": 3.937674790251698,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 9856
    },
    {
      "epoch": 3.9380743108270075,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 9857
    },
    {
      "epoch": 3.9384738314023173,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 9858
    },
    {
      "epoch": 3.9388733519776267,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 9859
    },
    {
      "epoch": 3.9392728725529365,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 9860
    },
    {
      "epoch": 3.9396723931282462,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 9861
    },
    {
      "epoch": 3.9400719137035556,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 9862
    },
    {
      "epoch": 3.9404714342788654,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 9863
    },
    {
      "epoch": 3.9408709548541747,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 9864
    },
    {
      "epoch": 3.9412704754294845,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 9865
    },
    {
      "epoch": 3.9416699960047943,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 9866
    },
    {
      "epoch": 3.942069516580104,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 9867
    },
    {
      "epoch": 3.9424690371554134,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 9868
    },
    {
      "epoch": 3.942868557730723,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 9869
    },
    {
      "epoch": 3.9432680783060325,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 9870
    },
    {
      "epoch": 3.9436675988813423,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 9871
    },
    {
      "epoch": 3.944067119456652,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 9872
    },
    {
      "epoch": 3.944466640031962,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 9873
    },
    {
      "epoch": 3.9448661606072712,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 9874
    },
    {
      "epoch": 3.945265681182581,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 9875
    },
    {
      "epoch": 3.9456652017578904,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 9876
    },
    {
      "epoch": 3.9460647223332,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 9877
    },
    {
      "epoch": 3.94646424290851,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 9878
    },
    {
      "epoch": 3.9468637634838193,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 9879
    },
    {
      "epoch": 3.947263284059129,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 9880
    },
    {
      "epoch": 3.9476628046344384,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 9881
    },
    {
      "epoch": 3.948062325209748,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 9882
    },
    {
      "epoch": 3.948461845785058,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 9883
    },
    {
      "epoch": 3.9488613663603678,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 9884
    },
    {
      "epoch": 3.949260886935677,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 9885
    },
    {
      "epoch": 3.949660407510987,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 9886
    },
    {
      "epoch": 3.9500599280862962,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 9887
    },
    {
      "epoch": 3.950459448661606,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 9888
    },
    {
      "epoch": 3.950858969236916,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 9889
    },
    {
      "epoch": 3.9512584898122256,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 9890
    },
    {
      "epoch": 3.951658010387535,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 9891
    },
    {
      "epoch": 3.9520575309628447,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 9892
    },
    {
      "epoch": 3.952457051538154,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 9893
    },
    {
      "epoch": 3.952856572113464,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 9894
    },
    {
      "epoch": 3.9532560926887736,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 9895
    },
    {
      "epoch": 3.953655613264083,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 9896
    },
    {
      "epoch": 3.9540551338393928,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 9897
    },
    {
      "epoch": 3.954454654414702,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 9898
    },
    {
      "epoch": 3.954854174990012,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 9899
    },
    {
      "epoch": 3.9552536955653217,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 9900
    },
    {
      "epoch": 3.9556532161406315,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 9901
    },
    {
      "epoch": 3.956052736715941,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 9902
    },
    {
      "epoch": 3.9564522572912506,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 9903
    },
    {
      "epoch": 3.95685177786656,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 9904
    },
    {
      "epoch": 3.9572512984418697,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 9905
    },
    {
      "epoch": 3.9576508190171795,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0748,
      "step": 9906
    },
    {
      "epoch": 3.958050339592489,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 9907
    },
    {
      "epoch": 3.9584498601677987,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 9908
    },
    {
      "epoch": 3.958849380743108,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 9909
    },
    {
      "epoch": 3.959248901318418,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 9910
    },
    {
      "epoch": 3.9596484218937276,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 9911
    },
    {
      "epoch": 3.9600479424690374,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 9912
    },
    {
      "epoch": 3.9604474630443467,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 9913
    },
    {
      "epoch": 3.9608469836196565,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 9914
    },
    {
      "epoch": 3.961246504194966,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 9915
    },
    {
      "epoch": 3.9616460247702756,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 9916
    },
    {
      "epoch": 3.9620455453455854,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 9917
    },
    {
      "epoch": 3.962445065920895,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 9918
    },
    {
      "epoch": 3.9628445864962045,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 9919
    },
    {
      "epoch": 3.9632441070715143,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 9920
    },
    {
      "epoch": 3.9636436276468237,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 9921
    },
    {
      "epoch": 3.9640431482221334,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 9922
    },
    {
      "epoch": 3.9644426687974432,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 9923
    },
    {
      "epoch": 3.9648421893727526,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 9924
    },
    {
      "epoch": 3.9652417099480624,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 9925
    },
    {
      "epoch": 3.9656412305233717,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 9926
    },
    {
      "epoch": 3.9660407510986815,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 9927
    },
    {
      "epoch": 3.9664402716739913,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 9928
    },
    {
      "epoch": 3.966839792249301,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 9929
    },
    {
      "epoch": 3.9672393128246104,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 9930
    },
    {
      "epoch": 3.96763883339992,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 9931
    },
    {
      "epoch": 3.9680383539752295,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 9932
    },
    {
      "epoch": 3.9684378745505393,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 9933
    },
    {
      "epoch": 3.968837395125849,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 9934
    },
    {
      "epoch": 3.969236915701159,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 9935
    },
    {
      "epoch": 3.9696364362764682,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 9936
    },
    {
      "epoch": 3.970035956851778,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 9937
    },
    {
      "epoch": 3.9704354774270874,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 9938
    },
    {
      "epoch": 3.970834998002397,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 9939
    },
    {
      "epoch": 3.971234518577707,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0636,
      "step": 9940
    },
    {
      "epoch": 3.9716340391530163,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 9941
    },
    {
      "epoch": 3.972033559728326,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 9942
    },
    {
      "epoch": 3.9724330803036354,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 9943
    },
    {
      "epoch": 3.972832600878945,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 9944
    },
    {
      "epoch": 3.973232121454255,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 9945
    },
    {
      "epoch": 3.9736316420295648,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 9946
    },
    {
      "epoch": 3.974031162604874,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 9947
    },
    {
      "epoch": 3.974430683180184,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 9948
    },
    {
      "epoch": 3.9748302037554932,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 9949
    },
    {
      "epoch": 3.975229724330803,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 9950
    },
    {
      "epoch": 3.975629244906113,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 9951
    },
    {
      "epoch": 3.976028765481422,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 9952
    },
    {
      "epoch": 3.976428286056732,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 9953
    },
    {
      "epoch": 3.9768278066320413,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 9954
    },
    {
      "epoch": 3.977227327207351,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 9955
    },
    {
      "epoch": 3.977626847782661,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0647,
      "step": 9956
    },
    {
      "epoch": 3.9780263683579706,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 9957
    },
    {
      "epoch": 3.97842588893328,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 9958
    },
    {
      "epoch": 3.9788254095085898,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 9959
    },
    {
      "epoch": 3.979224930083899,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 9960
    },
    {
      "epoch": 3.979624450659209,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 9961
    },
    {
      "epoch": 3.9800239712345187,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 9962
    },
    {
      "epoch": 3.9804234918098285,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 9963
    },
    {
      "epoch": 3.980823012385138,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 9964
    },
    {
      "epoch": 3.9812225329604476,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 9965
    },
    {
      "epoch": 3.981622053535757,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 9966
    },
    {
      "epoch": 3.9820215741110667,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 9967
    },
    {
      "epoch": 3.9824210946863765,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 9968
    },
    {
      "epoch": 3.982820615261686,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 9969
    },
    {
      "epoch": 3.9832201358369956,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 9970
    },
    {
      "epoch": 3.983619656412305,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 9971
    },
    {
      "epoch": 3.9840191769876148,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 9972
    },
    {
      "epoch": 3.9844186975629245,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 9973
    },
    {
      "epoch": 3.9848182181382343,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 9974
    },
    {
      "epoch": 3.9852177387135437,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 9975
    },
    {
      "epoch": 3.9856172592888535,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 9976
    },
    {
      "epoch": 3.986016779864163,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 9977
    },
    {
      "epoch": 3.9864163004394726,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 9978
    },
    {
      "epoch": 3.9868158210147824,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 9979
    },
    {
      "epoch": 3.987215341590092,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 9980
    },
    {
      "epoch": 3.9876148621654015,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 9981
    },
    {
      "epoch": 3.9880143827407113,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 9982
    },
    {
      "epoch": 3.9884139033160206,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 9983
    },
    {
      "epoch": 3.9888134238913304,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 9984
    },
    {
      "epoch": 3.98921294446664,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 9985
    },
    {
      "epoch": 3.9896124650419496,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 9986
    },
    {
      "epoch": 3.9900119856172593,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 9987
    },
    {
      "epoch": 3.9904115061925687,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 9988
    },
    {
      "epoch": 3.9908110267678785,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 9989
    },
    {
      "epoch": 3.9912105473431883,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0695,
      "step": 9990
    },
    {
      "epoch": 3.991610067918498,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 9991
    },
    {
      "epoch": 3.9920095884938074,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 9992
    },
    {
      "epoch": 3.992409109069117,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 9993
    },
    {
      "epoch": 3.9928086296444265,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 9994
    },
    {
      "epoch": 3.9932081502197363,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 9995
    },
    {
      "epoch": 3.993607670795046,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 9996
    },
    {
      "epoch": 3.9940071913703554,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 9997
    },
    {
      "epoch": 3.994406711945665,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 9998
    },
    {
      "epoch": 3.9948062325209746,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 9999
    },
    {
      "epoch": 3.9952057530962843,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 10000
    },
    {
      "epoch": 3.995605273671594,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 10001
    },
    {
      "epoch": 3.996004794246904,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 10002
    },
    {
      "epoch": 3.9964043148222133,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 10003
    },
    {
      "epoch": 3.996803835397523,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 10004
    },
    {
      "epoch": 3.9972033559728324,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 10005
    },
    {
      "epoch": 3.997602876548142,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 10006
    },
    {
      "epoch": 3.998002397123452,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 10007
    },
    {
      "epoch": 3.9984019176987617,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 10008
    },
    {
      "epoch": 3.998801438274071,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 10009
    },
    {
      "epoch": 3.999200958849381,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 10010
    },
    {
      "epoch": 3.99960047942469,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0619,
      "step": 10011
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 10012
    },
    {
      "epoch": 4.00039952057531,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 10013
    },
    {
      "epoch": 4.00079904115062,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 10014
    },
    {
      "epoch": 4.0011985617259285,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 10015
    },
    {
      "epoch": 4.001598082301238,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 10016
    },
    {
      "epoch": 4.001997602876548,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 10017
    },
    {
      "epoch": 4.002397123451858,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 10018
    },
    {
      "epoch": 4.002796644027168,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 10019
    },
    {
      "epoch": 4.003196164602477,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 10020
    },
    {
      "epoch": 4.003595685177786,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 10021
    },
    {
      "epoch": 4.003995205753096,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 10022
    },
    {
      "epoch": 4.004394726328406,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 10023
    },
    {
      "epoch": 4.004794246903716,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 10024
    },
    {
      "epoch": 4.0051937674790254,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 10025
    },
    {
      "epoch": 4.005593288054335,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 10026
    },
    {
      "epoch": 4.005992808629644,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 10027
    },
    {
      "epoch": 4.006392329204954,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 10028
    },
    {
      "epoch": 4.006791849780264,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 10029
    },
    {
      "epoch": 4.0071913703555735,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 10030
    },
    {
      "epoch": 4.007590890930883,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 10031
    },
    {
      "epoch": 4.007990411506192,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 10032
    },
    {
      "epoch": 4.008389932081502,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 10033
    },
    {
      "epoch": 4.008789452656812,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 10034
    },
    {
      "epoch": 4.0091889732321215,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 10035
    },
    {
      "epoch": 4.009588493807431,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 10036
    },
    {
      "epoch": 4.009988014382741,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 10037
    },
    {
      "epoch": 4.01038753495805,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 10038
    },
    {
      "epoch": 4.01078705553336,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 10039
    },
    {
      "epoch": 4.01118657610867,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 10040
    },
    {
      "epoch": 4.011586096683979,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 10041
    },
    {
      "epoch": 4.011985617259289,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 10042
    },
    {
      "epoch": 4.012385137834598,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 10043
    },
    {
      "epoch": 4.012784658409908,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 10044
    },
    {
      "epoch": 4.013184178985218,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 10045
    },
    {
      "epoch": 4.013583699560527,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 10046
    },
    {
      "epoch": 4.013983220135837,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 10047
    },
    {
      "epoch": 4.014382740711147,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 10048
    },
    {
      "epoch": 4.014782261286456,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 10049
    },
    {
      "epoch": 4.015181781861766,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 10050
    },
    {
      "epoch": 4.0155813024370755,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 10051
    },
    {
      "epoch": 4.015980823012385,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 10052
    },
    {
      "epoch": 4.016380343587695,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 10053
    },
    {
      "epoch": 4.016779864163005,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 10054
    },
    {
      "epoch": 4.017179384738314,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 10055
    },
    {
      "epoch": 4.0175789053136235,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 10056
    },
    {
      "epoch": 4.017978425888933,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 10057
    },
    {
      "epoch": 4.018377946464243,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 10058
    },
    {
      "epoch": 4.018777467039553,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 10059
    },
    {
      "epoch": 4.019176987614862,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 10060
    },
    {
      "epoch": 4.0195765081901715,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 10061
    },
    {
      "epoch": 4.019976028765481,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 10062
    },
    {
      "epoch": 4.020375549340791,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 10063
    },
    {
      "epoch": 4.020775069916101,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 10064
    },
    {
      "epoch": 4.021174590491411,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 10065
    },
    {
      "epoch": 4.02157411106672,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 10066
    },
    {
      "epoch": 4.021973631642029,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 10067
    },
    {
      "epoch": 4.022373152217339,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 10068
    },
    {
      "epoch": 4.022772672792649,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 10069
    },
    {
      "epoch": 4.023172193367959,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 10070
    },
    {
      "epoch": 4.0235717139432685,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 10071
    },
    {
      "epoch": 4.023971234518577,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 10072
    },
    {
      "epoch": 4.024370755093887,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 10073
    },
    {
      "epoch": 4.024770275669197,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 10074
    },
    {
      "epoch": 4.025169796244507,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 10075
    },
    {
      "epoch": 4.025569316819817,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 10076
    },
    {
      "epoch": 4.0259688373951255,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 10077
    },
    {
      "epoch": 4.026368357970435,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 10078
    },
    {
      "epoch": 4.026767878545745,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 10079
    },
    {
      "epoch": 4.027167399121055,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 10080
    },
    {
      "epoch": 4.027566919696365,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 10081
    },
    {
      "epoch": 4.027966440271674,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 10082
    },
    {
      "epoch": 4.028365960846983,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 10083
    },
    {
      "epoch": 4.028765481422293,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 10084
    },
    {
      "epoch": 4.029165001997603,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 10085
    },
    {
      "epoch": 4.029564522572913,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 10086
    },
    {
      "epoch": 4.029964043148222,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 10087
    },
    {
      "epoch": 4.030363563723531,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 10088
    },
    {
      "epoch": 4.030763084298841,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 10089
    },
    {
      "epoch": 4.031162604874151,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 10090
    },
    {
      "epoch": 4.031562125449461,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 10091
    },
    {
      "epoch": 4.0319616460247705,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 10092
    },
    {
      "epoch": 4.03236116660008,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 10093
    },
    {
      "epoch": 4.032760687175389,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 10094
    },
    {
      "epoch": 4.033160207750699,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 10095
    },
    {
      "epoch": 4.033559728326009,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 10096
    },
    {
      "epoch": 4.0339592489013185,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 10097
    },
    {
      "epoch": 4.034358769476628,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 10098
    },
    {
      "epoch": 4.034758290051938,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 10099
    },
    {
      "epoch": 4.035157810627247,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 10100
    },
    {
      "epoch": 4.035557331202557,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 10101
    },
    {
      "epoch": 4.035956851777867,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 10102
    },
    {
      "epoch": 4.036356372353176,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 10103
    },
    {
      "epoch": 4.036755892928486,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 10104
    },
    {
      "epoch": 4.037155413503795,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 10105
    },
    {
      "epoch": 4.037554934079105,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 10106
    },
    {
      "epoch": 4.037954454654415,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 10107
    },
    {
      "epoch": 4.038353975229724,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 10108
    },
    {
      "epoch": 4.038753495805034,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 10109
    },
    {
      "epoch": 4.039153016380344,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 10110
    },
    {
      "epoch": 4.039552536955653,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 10111
    },
    {
      "epoch": 4.039952057530963,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 10112
    },
    {
      "epoch": 4.040351578106272,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 10113
    },
    {
      "epoch": 4.040751098681582,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 10114
    },
    {
      "epoch": 4.041150619256892,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 10115
    },
    {
      "epoch": 4.041550139832202,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 10116
    },
    {
      "epoch": 4.041949660407511,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 10117
    },
    {
      "epoch": 4.0423491809828205,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 10118
    },
    {
      "epoch": 4.04274870155813,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 10119
    },
    {
      "epoch": 4.04314822213344,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 10120
    },
    {
      "epoch": 4.04354774270875,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 10121
    },
    {
      "epoch": 4.043947263284059,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 10122
    },
    {
      "epoch": 4.0443467838593685,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 10123
    },
    {
      "epoch": 4.044746304434678,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 10124
    },
    {
      "epoch": 4.045145825009988,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 10125
    },
    {
      "epoch": 4.045545345585298,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 10126
    },
    {
      "epoch": 4.045944866160608,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 10127
    },
    {
      "epoch": 4.046344386735917,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 10128
    },
    {
      "epoch": 4.046743907311226,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 10129
    },
    {
      "epoch": 4.047143427886536,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 10130
    },
    {
      "epoch": 4.047542948461846,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 10131
    },
    {
      "epoch": 4.047942469037156,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 10132
    },
    {
      "epoch": 4.048341989612465,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 10133
    },
    {
      "epoch": 4.048741510187774,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 10134
    },
    {
      "epoch": 4.049141030763084,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 10135
    },
    {
      "epoch": 4.049540551338394,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 10136
    },
    {
      "epoch": 4.049940071913704,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 10137
    },
    {
      "epoch": 4.0503395924890135,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 10138
    },
    {
      "epoch": 4.050739113064322,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 10139
    },
    {
      "epoch": 4.051138633639632,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 10140
    },
    {
      "epoch": 4.051538154214942,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 10141
    },
    {
      "epoch": 4.051937674790252,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 10142
    },
    {
      "epoch": 4.052337195365562,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 10143
    },
    {
      "epoch": 4.052736715940871,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 10144
    },
    {
      "epoch": 4.05313623651618,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 10145
    },
    {
      "epoch": 4.05353575709149,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 10146
    },
    {
      "epoch": 4.0539352776668,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 10147
    },
    {
      "epoch": 4.05433479824211,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 10148
    },
    {
      "epoch": 4.054734318817419,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 10149
    },
    {
      "epoch": 4.055133839392728,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 10150
    },
    {
      "epoch": 4.055533359968038,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 10151
    },
    {
      "epoch": 4.055932880543348,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 10152
    },
    {
      "epoch": 4.056332401118658,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 10153
    },
    {
      "epoch": 4.0567319216939675,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 10154
    },
    {
      "epoch": 4.057131442269277,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 10155
    },
    {
      "epoch": 4.057530962844586,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 10156
    },
    {
      "epoch": 4.057930483419896,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 10157
    },
    {
      "epoch": 4.058330003995206,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 10158
    },
    {
      "epoch": 4.0587295245705155,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 10159
    },
    {
      "epoch": 4.059129045145825,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 10160
    },
    {
      "epoch": 4.059528565721135,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 10161
    },
    {
      "epoch": 4.059928086296444,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 10162
    },
    {
      "epoch": 4.060327606871754,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 10163
    },
    {
      "epoch": 4.0607271274470635,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 10164
    },
    {
      "epoch": 4.061126648022373,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 10165
    },
    {
      "epoch": 4.061526168597683,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 10166
    },
    {
      "epoch": 4.061925689172992,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 10167
    },
    {
      "epoch": 4.062325209748302,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 10168
    },
    {
      "epoch": 4.062724730323612,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 10169
    },
    {
      "epoch": 4.063124250898921,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 10170
    },
    {
      "epoch": 4.063523771474231,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 10171
    },
    {
      "epoch": 4.063923292049541,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 10172
    },
    {
      "epoch": 4.06432281262485,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 10173
    },
    {
      "epoch": 4.06472233320016,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 10174
    },
    {
      "epoch": 4.065121853775469,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 10175
    },
    {
      "epoch": 4.065521374350779,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 10176
    },
    {
      "epoch": 4.065920894926089,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 10177
    },
    {
      "epoch": 4.066320415501398,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 10178
    },
    {
      "epoch": 4.066719936076708,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 10179
    },
    {
      "epoch": 4.0671194566520175,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 10180
    },
    {
      "epoch": 4.067518977227327,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 10181
    },
    {
      "epoch": 4.067918497802637,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 10182
    },
    {
      "epoch": 4.068318018377947,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 10183
    },
    {
      "epoch": 4.068717538953256,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 10184
    },
    {
      "epoch": 4.0691170595285655,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 10185
    },
    {
      "epoch": 4.069516580103875,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 10186
    },
    {
      "epoch": 4.069916100679185,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 10187
    },
    {
      "epoch": 4.070315621254495,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 10188
    },
    {
      "epoch": 4.070715141829805,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 10189
    },
    {
      "epoch": 4.0711146624051135,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 10190
    },
    {
      "epoch": 4.071514182980423,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 10191
    },
    {
      "epoch": 4.071913703555733,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 10192
    },
    {
      "epoch": 4.072313224131043,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 10193
    },
    {
      "epoch": 4.072712744706353,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 10194
    },
    {
      "epoch": 4.073112265281662,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 10195
    },
    {
      "epoch": 4.073511785856971,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 10196
    },
    {
      "epoch": 4.073911306432281,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 10197
    },
    {
      "epoch": 4.074310827007591,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 10198
    },
    {
      "epoch": 4.074710347582901,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 10199
    },
    {
      "epoch": 4.0751098681582105,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 10200
    },
    {
      "epoch": 4.075509388733519,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 10201
    },
    {
      "epoch": 4.075908909308829,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 10202
    },
    {
      "epoch": 4.076308429884139,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 10203
    },
    {
      "epoch": 4.076707950459449,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0701,
      "step": 10204
    },
    {
      "epoch": 4.077107471034759,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 10205
    },
    {
      "epoch": 4.077506991610068,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 10206
    },
    {
      "epoch": 4.077906512185377,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 10207
    },
    {
      "epoch": 4.078306032760687,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 10208
    },
    {
      "epoch": 4.078705553335997,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 10209
    },
    {
      "epoch": 4.079105073911307,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 10210
    },
    {
      "epoch": 4.079504594486616,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 10211
    },
    {
      "epoch": 4.079904115061925,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 10212
    },
    {
      "epoch": 4.080303635637235,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 10213
    },
    {
      "epoch": 4.080703156212545,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 10214
    },
    {
      "epoch": 4.081102676787855,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 10215
    },
    {
      "epoch": 4.081502197363164,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 10216
    },
    {
      "epoch": 4.081901717938474,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 10217
    },
    {
      "epoch": 4.082301238513783,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 10218
    },
    {
      "epoch": 4.082700759089093,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 10219
    },
    {
      "epoch": 4.083100279664403,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 10220
    },
    {
      "epoch": 4.0834998002397125,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 10221
    },
    {
      "epoch": 4.083899320815022,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 10222
    },
    {
      "epoch": 4.084298841390331,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 10223
    },
    {
      "epoch": 4.084698361965641,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 10224
    },
    {
      "epoch": 4.085097882540951,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 10225
    },
    {
      "epoch": 4.0854974031162605,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 10226
    },
    {
      "epoch": 4.08589692369157,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 10227
    },
    {
      "epoch": 4.08629644426688,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 10228
    },
    {
      "epoch": 4.086695964842189,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 10229
    },
    {
      "epoch": 4.087095485417499,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 10230
    },
    {
      "epoch": 4.087495005992809,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 10231
    },
    {
      "epoch": 4.087894526568118,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 10232
    },
    {
      "epoch": 4.088294047143428,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 10233
    },
    {
      "epoch": 4.088693567718738,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 10234
    },
    {
      "epoch": 4.089093088294047,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 10235
    },
    {
      "epoch": 4.089492608869357,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 10236
    },
    {
      "epoch": 4.089892129444666,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 10237
    },
    {
      "epoch": 4.090291650019976,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 10238
    },
    {
      "epoch": 4.090691170595286,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 10239
    },
    {
      "epoch": 4.091090691170595,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 10240
    },
    {
      "epoch": 4.091490211745905,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 10241
    },
    {
      "epoch": 4.0918897323212144,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 10242
    },
    {
      "epoch": 4.092289252896524,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 10243
    },
    {
      "epoch": 4.092688773471834,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 10244
    },
    {
      "epoch": 4.093088294047144,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 10245
    },
    {
      "epoch": 4.093487814622453,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 10246
    },
    {
      "epoch": 4.0938873351977625,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 10247
    },
    {
      "epoch": 4.094286855773072,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0634,
      "step": 10248
    },
    {
      "epoch": 4.094686376348382,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 10249
    },
    {
      "epoch": 4.095085896923692,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 10250
    },
    {
      "epoch": 4.095485417499002,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 10251
    },
    {
      "epoch": 4.0958849380743105,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 10252
    },
    {
      "epoch": 4.09628445864962,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 10253
    },
    {
      "epoch": 4.09668397922493,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 10254
    },
    {
      "epoch": 4.09708349980024,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 10255
    },
    {
      "epoch": 4.09748302037555,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 10256
    },
    {
      "epoch": 4.097882540950859,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 10257
    },
    {
      "epoch": 4.098282061526168,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 10258
    },
    {
      "epoch": 4.098681582101478,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 10259
    },
    {
      "epoch": 4.099081102676788,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 10260
    },
    {
      "epoch": 4.099480623252098,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 10261
    },
    {
      "epoch": 4.0998801438274075,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 10262
    },
    {
      "epoch": 4.100279664402716,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 10263
    },
    {
      "epoch": 4.100679184978026,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 10264
    },
    {
      "epoch": 4.101078705553336,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 10265
    },
    {
      "epoch": 4.101478226128646,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 10266
    },
    {
      "epoch": 4.1018777467039556,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 10267
    },
    {
      "epoch": 4.1022772672792644,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 10268
    },
    {
      "epoch": 4.102676787854574,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 10269
    },
    {
      "epoch": 4.103076308429884,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 10270
    },
    {
      "epoch": 4.103475829005194,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 10271
    },
    {
      "epoch": 4.103875349580504,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 10272
    },
    {
      "epoch": 4.104274870155813,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 10273
    },
    {
      "epoch": 4.104674390731122,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 10274
    },
    {
      "epoch": 4.105073911306432,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 10275
    },
    {
      "epoch": 4.105473431881742,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 10276
    },
    {
      "epoch": 4.105872952457052,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 10277
    },
    {
      "epoch": 4.106272473032361,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 10278
    },
    {
      "epoch": 4.106671993607671,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 10279
    },
    {
      "epoch": 4.10707151418298,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 10280
    },
    {
      "epoch": 4.10747103475829,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 10281
    },
    {
      "epoch": 4.1078705553336,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 10282
    },
    {
      "epoch": 4.1082700759089095,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 10283
    },
    {
      "epoch": 4.108669596484219,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 10284
    },
    {
      "epoch": 4.109069117059528,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 10285
    },
    {
      "epoch": 4.109468637634838,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 10286
    },
    {
      "epoch": 4.109868158210148,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 10287
    },
    {
      "epoch": 4.1102676787854575,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 10288
    },
    {
      "epoch": 4.110667199360767,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 10289
    },
    {
      "epoch": 4.111066719936077,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 10290
    },
    {
      "epoch": 4.111466240511386,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 10291
    },
    {
      "epoch": 4.111865761086696,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 10292
    },
    {
      "epoch": 4.112265281662006,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 10293
    },
    {
      "epoch": 4.112664802237315,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 10294
    },
    {
      "epoch": 4.113064322812625,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 10295
    },
    {
      "epoch": 4.113463843387935,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 10296
    },
    {
      "epoch": 4.113863363963244,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 10297
    },
    {
      "epoch": 4.114262884538554,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 10298
    },
    {
      "epoch": 4.114662405113863,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 10299
    },
    {
      "epoch": 4.115061925689173,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 10300
    },
    {
      "epoch": 4.115461446264483,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 10301
    },
    {
      "epoch": 4.115860966839792,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 10302
    },
    {
      "epoch": 4.116260487415102,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 10303
    },
    {
      "epoch": 4.116660007990411,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 10304
    },
    {
      "epoch": 4.117059528565721,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 10305
    },
    {
      "epoch": 4.117459049141031,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 10306
    },
    {
      "epoch": 4.117858569716341,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 10307
    },
    {
      "epoch": 4.11825809029165,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 10308
    },
    {
      "epoch": 4.1186576108669595,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 10309
    },
    {
      "epoch": 4.119057131442269,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 10310
    },
    {
      "epoch": 4.119456652017579,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0618,
      "step": 10311
    },
    {
      "epoch": 4.119856172592889,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 10312
    },
    {
      "epoch": 4.120255693168198,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 10313
    },
    {
      "epoch": 4.1206552137435075,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 10314
    },
    {
      "epoch": 4.121054734318817,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 10315
    },
    {
      "epoch": 4.121454254894127,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 10316
    },
    {
      "epoch": 4.121853775469437,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 10317
    },
    {
      "epoch": 4.122253296044747,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 10318
    },
    {
      "epoch": 4.122652816620056,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 10319
    },
    {
      "epoch": 4.123052337195365,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 10320
    },
    {
      "epoch": 4.123451857770675,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 10321
    },
    {
      "epoch": 4.123851378345985,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 10322
    },
    {
      "epoch": 4.124250898921295,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 10323
    },
    {
      "epoch": 4.1246504194966045,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 10324
    },
    {
      "epoch": 4.125049940071913,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 10325
    },
    {
      "epoch": 4.125449460647223,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 10326
    },
    {
      "epoch": 4.125848981222533,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 10327
    },
    {
      "epoch": 4.126248501797843,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 10328
    },
    {
      "epoch": 4.1266480223731525,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 10329
    },
    {
      "epoch": 4.127047542948461,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 10330
    },
    {
      "epoch": 4.127447063523771,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 10331
    },
    {
      "epoch": 4.127846584099081,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 10332
    },
    {
      "epoch": 4.128246104674391,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 10333
    },
    {
      "epoch": 4.128645625249701,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 10334
    },
    {
      "epoch": 4.12904514582501,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 10335
    },
    {
      "epoch": 4.129444666400319,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 10336
    },
    {
      "epoch": 4.129844186975629,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 10337
    },
    {
      "epoch": 4.130243707550939,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 10338
    },
    {
      "epoch": 4.130643228126249,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 10339
    },
    {
      "epoch": 4.131042748701558,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 10340
    },
    {
      "epoch": 4.131442269276867,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 10341
    },
    {
      "epoch": 4.131841789852177,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 10342
    },
    {
      "epoch": 4.132241310427487,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 10343
    },
    {
      "epoch": 4.132640831002797,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 10344
    },
    {
      "epoch": 4.1330403515781065,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 10345
    },
    {
      "epoch": 4.133439872153416,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 10346
    },
    {
      "epoch": 4.133839392728725,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 10347
    },
    {
      "epoch": 4.134238913304035,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 10348
    },
    {
      "epoch": 4.134638433879345,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 10349
    },
    {
      "epoch": 4.1350379544546545,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 10350
    },
    {
      "epoch": 4.135437475029964,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.067,
      "step": 10351
    },
    {
      "epoch": 4.135836995605274,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 10352
    },
    {
      "epoch": 4.136236516180583,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 10353
    },
    {
      "epoch": 4.136636036755893,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 10354
    },
    {
      "epoch": 4.1370355573312025,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 10355
    },
    {
      "epoch": 4.137435077906512,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 10356
    },
    {
      "epoch": 4.137834598481822,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 10357
    },
    {
      "epoch": 4.138234119057131,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 10358
    },
    {
      "epoch": 4.138633639632441,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 10359
    },
    {
      "epoch": 4.139033160207751,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 10360
    },
    {
      "epoch": 4.13943268078306,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 10361
    },
    {
      "epoch": 4.13983220135837,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 10362
    },
    {
      "epoch": 4.14023172193368,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 10363
    },
    {
      "epoch": 4.140631242508989,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 10364
    },
    {
      "epoch": 4.141030763084299,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 10365
    },
    {
      "epoch": 4.141430283659608,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 10366
    },
    {
      "epoch": 4.141829804234918,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 10367
    },
    {
      "epoch": 4.142229324810228,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 10368
    },
    {
      "epoch": 4.142628845385538,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 10369
    },
    {
      "epoch": 4.143028365960847,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 10370
    },
    {
      "epoch": 4.1434278865361565,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 10371
    },
    {
      "epoch": 4.143827407111466,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 10372
    },
    {
      "epoch": 4.144226927686776,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 10373
    },
    {
      "epoch": 4.144626448262086,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 10374
    },
    {
      "epoch": 4.145025968837395,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 10375
    },
    {
      "epoch": 4.1454254894127045,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 10376
    },
    {
      "epoch": 4.145825009988014,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 10377
    },
    {
      "epoch": 4.146224530563324,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 10378
    },
    {
      "epoch": 4.146624051138634,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 10379
    },
    {
      "epoch": 4.147023571713944,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 10380
    },
    {
      "epoch": 4.1474230922892525,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 10381
    },
    {
      "epoch": 4.147822612864562,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 10382
    },
    {
      "epoch": 4.148222133439872,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 10383
    },
    {
      "epoch": 4.148621654015182,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 10384
    },
    {
      "epoch": 4.149021174590492,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 10385
    },
    {
      "epoch": 4.1494206951658015,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 10386
    },
    {
      "epoch": 4.14982021574111,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 10387
    },
    {
      "epoch": 4.15021973631642,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 10388
    },
    {
      "epoch": 4.15061925689173,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 10389
    },
    {
      "epoch": 4.15101877746704,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 10390
    },
    {
      "epoch": 4.1514182980423495,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 10391
    },
    {
      "epoch": 4.151817818617658,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 10392
    },
    {
      "epoch": 4.152217339192968,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 10393
    },
    {
      "epoch": 4.152616859768278,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 10394
    },
    {
      "epoch": 4.153016380343588,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 10395
    },
    {
      "epoch": 4.153415900918898,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 10396
    },
    {
      "epoch": 4.153815421494207,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 10397
    },
    {
      "epoch": 4.154214942069516,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 10398
    },
    {
      "epoch": 4.154614462644826,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 10399
    },
    {
      "epoch": 4.155013983220136,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 10400
    },
    {
      "epoch": 4.155413503795446,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 10401
    },
    {
      "epoch": 4.155813024370755,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 10402
    },
    {
      "epoch": 4.156212544946065,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 10403
    },
    {
      "epoch": 4.156612065521374,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 10404
    },
    {
      "epoch": 4.157011586096684,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 10405
    },
    {
      "epoch": 4.157411106671994,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 10406
    },
    {
      "epoch": 4.157810627247303,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 10407
    },
    {
      "epoch": 4.158210147822613,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 10408
    },
    {
      "epoch": 4.158609668397922,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 10409
    },
    {
      "epoch": 4.159009188973232,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 10410
    },
    {
      "epoch": 4.159408709548542,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 10411
    },
    {
      "epoch": 4.1598082301238515,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 10412
    },
    {
      "epoch": 4.160207750699161,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 10413
    },
    {
      "epoch": 4.160607271274471,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 10414
    },
    {
      "epoch": 4.16100679184978,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 10415
    },
    {
      "epoch": 4.16140631242509,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 10416
    },
    {
      "epoch": 4.1618058330003995,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 10417
    },
    {
      "epoch": 4.162205353575709,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 10418
    },
    {
      "epoch": 4.162604874151019,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 10419
    },
    {
      "epoch": 4.163004394726328,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 10420
    },
    {
      "epoch": 4.163403915301638,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 10421
    },
    {
      "epoch": 4.163803435876948,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 10422
    },
    {
      "epoch": 4.164202956452257,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 10423
    },
    {
      "epoch": 4.164602477027567,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 10424
    },
    {
      "epoch": 4.165001997602877,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 10425
    },
    {
      "epoch": 4.165401518178186,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 10426
    },
    {
      "epoch": 4.165801038753496,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 10427
    },
    {
      "epoch": 4.166200559328805,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 10428
    },
    {
      "epoch": 4.166600079904115,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 10429
    },
    {
      "epoch": 4.166999600479425,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 10430
    },
    {
      "epoch": 4.167399121054734,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 10431
    },
    {
      "epoch": 4.167798641630044,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 10432
    },
    {
      "epoch": 4.168198162205353,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 10433
    },
    {
      "epoch": 4.168597682780663,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 10434
    },
    {
      "epoch": 4.168997203355973,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 10435
    },
    {
      "epoch": 4.169396723931283,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 10436
    },
    {
      "epoch": 4.169796244506592,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 10437
    },
    {
      "epoch": 4.1701957650819015,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 10438
    },
    {
      "epoch": 4.170595285657211,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 10439
    },
    {
      "epoch": 4.170994806232521,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 10440
    },
    {
      "epoch": 4.171394326807831,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 10441
    },
    {
      "epoch": 4.171793847383141,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 10442
    },
    {
      "epoch": 4.1721933679584495,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 10443
    },
    {
      "epoch": 4.172592888533759,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 10444
    },
    {
      "epoch": 4.172992409109069,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 10445
    },
    {
      "epoch": 4.173391929684379,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 10446
    },
    {
      "epoch": 4.173791450259689,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 10447
    },
    {
      "epoch": 4.174190970834998,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 10448
    },
    {
      "epoch": 4.174590491410307,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 10449
    },
    {
      "epoch": 4.174990011985617,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 10450
    },
    {
      "epoch": 4.175389532560927,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 10451
    },
    {
      "epoch": 4.175789053136237,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 10452
    },
    {
      "epoch": 4.1761885737115465,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 10453
    },
    {
      "epoch": 4.176588094286855,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 10454
    },
    {
      "epoch": 4.176987614862165,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 10455
    },
    {
      "epoch": 4.177387135437475,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 10456
    },
    {
      "epoch": 4.177786656012785,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 10457
    },
    {
      "epoch": 4.1781861765880945,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 10458
    },
    {
      "epoch": 4.178585697163404,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 10459
    },
    {
      "epoch": 4.178985217738713,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 10460
    },
    {
      "epoch": 4.179384738314023,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 10461
    },
    {
      "epoch": 4.179784258889333,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 10462
    },
    {
      "epoch": 4.180183779464643,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 10463
    },
    {
      "epoch": 4.180583300039952,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 10464
    },
    {
      "epoch": 4.180982820615261,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 10465
    },
    {
      "epoch": 4.181382341190571,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 10466
    },
    {
      "epoch": 4.181781861765881,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 10467
    },
    {
      "epoch": 4.182181382341191,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 10468
    },
    {
      "epoch": 4.1825809029165,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 10469
    },
    {
      "epoch": 4.18298042349181,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 10470
    },
    {
      "epoch": 4.183379944067119,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 10471
    },
    {
      "epoch": 4.183779464642429,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 10472
    },
    {
      "epoch": 4.184178985217739,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 10473
    },
    {
      "epoch": 4.1845785057930485,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 10474
    },
    {
      "epoch": 4.184978026368358,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 10475
    },
    {
      "epoch": 4.185377546943668,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 10476
    },
    {
      "epoch": 4.185777067518977,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 10477
    },
    {
      "epoch": 4.186176588094287,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 10478
    },
    {
      "epoch": 4.1865761086695965,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 10479
    },
    {
      "epoch": 4.186975629244906,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 10480
    },
    {
      "epoch": 4.187375149820216,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 10481
    },
    {
      "epoch": 4.187774670395525,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 10482
    },
    {
      "epoch": 4.188174190970835,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 10483
    },
    {
      "epoch": 4.1885737115461446,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 10484
    },
    {
      "epoch": 4.188973232121454,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 10485
    },
    {
      "epoch": 4.189372752696764,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 10486
    },
    {
      "epoch": 4.189772273272074,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 10487
    },
    {
      "epoch": 4.190171793847383,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 10488
    },
    {
      "epoch": 4.190571314422693,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 10489
    },
    {
      "epoch": 4.190970834998002,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 10490
    },
    {
      "epoch": 4.191370355573312,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 10491
    },
    {
      "epoch": 4.191769876148622,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 10492
    },
    {
      "epoch": 4.192169396723932,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 10493
    },
    {
      "epoch": 4.192568917299241,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 10494
    },
    {
      "epoch": 4.19296843787455,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 10495
    },
    {
      "epoch": 4.19336795844986,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 10496
    },
    {
      "epoch": 4.19376747902517,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0625,
      "step": 10497
    },
    {
      "epoch": 4.19416699960048,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 10498
    },
    {
      "epoch": 4.194566520175789,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 10499
    },
    {
      "epoch": 4.1949660407510985,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 10500
    },
    {
      "epoch": 4.195365561326408,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 10501
    },
    {
      "epoch": 4.195765081901718,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 10502
    },
    {
      "epoch": 4.196164602477028,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 10503
    },
    {
      "epoch": 4.196564123052338,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 10504
    },
    {
      "epoch": 4.1969636436276465,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 10505
    },
    {
      "epoch": 4.197363164202956,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 10506
    },
    {
      "epoch": 4.197762684778266,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 10507
    },
    {
      "epoch": 4.198162205353576,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0628,
      "step": 10508
    },
    {
      "epoch": 4.198561725928886,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 10509
    },
    {
      "epoch": 4.1989612465041946,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 10510
    },
    {
      "epoch": 4.199360767079504,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 10511
    },
    {
      "epoch": 4.199760287654814,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 10512
    },
    {
      "epoch": 4.200159808230124,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 10513
    },
    {
      "epoch": 4.200559328805434,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 10514
    },
    {
      "epoch": 4.2009588493807435,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 10515
    },
    {
      "epoch": 4.201358369956052,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 10516
    },
    {
      "epoch": 4.201757890531362,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 10517
    },
    {
      "epoch": 4.202157411106672,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 10518
    },
    {
      "epoch": 4.202556931681982,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 10519
    },
    {
      "epoch": 4.2029564522572915,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 10520
    },
    {
      "epoch": 4.2033559728326,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 10521
    },
    {
      "epoch": 4.20375549340791,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 10522
    },
    {
      "epoch": 4.20415501398322,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 10523
    },
    {
      "epoch": 4.20455453455853,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 10524
    },
    {
      "epoch": 4.20495405513384,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 10525
    },
    {
      "epoch": 4.205353575709149,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 10526
    },
    {
      "epoch": 4.205753096284458,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 10527
    },
    {
      "epoch": 4.206152616859768,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 10528
    },
    {
      "epoch": 4.206552137435078,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 10529
    },
    {
      "epoch": 4.206951658010388,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 10530
    },
    {
      "epoch": 4.207351178585697,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 10531
    },
    {
      "epoch": 4.207750699161007,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 10532
    },
    {
      "epoch": 4.208150219736316,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 10533
    },
    {
      "epoch": 4.208549740311626,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 10534
    },
    {
      "epoch": 4.208949260886936,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 10535
    },
    {
      "epoch": 4.2093487814622454,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 10536
    },
    {
      "epoch": 4.209748302037555,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 10537
    },
    {
      "epoch": 4.210147822612864,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 10538
    },
    {
      "epoch": 4.210547343188174,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 10539
    },
    {
      "epoch": 4.210946863763484,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 10540
    },
    {
      "epoch": 4.2113463843387935,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 10541
    },
    {
      "epoch": 4.211745904914103,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 10542
    },
    {
      "epoch": 4.212145425489413,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 10543
    },
    {
      "epoch": 4.212544946064722,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 10544
    },
    {
      "epoch": 4.212944466640032,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 10545
    },
    {
      "epoch": 4.2133439872153415,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 10546
    },
    {
      "epoch": 4.213743507790651,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 10547
    },
    {
      "epoch": 4.214143028365961,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 10548
    },
    {
      "epoch": 4.214542548941271,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 10549
    },
    {
      "epoch": 4.21494206951658,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 10550
    },
    {
      "epoch": 4.21534159009189,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 10551
    },
    {
      "epoch": 4.215741110667199,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 10552
    },
    {
      "epoch": 4.216140631242509,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 10553
    },
    {
      "epoch": 4.216540151817819,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 10554
    },
    {
      "epoch": 4.216939672393128,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 10555
    },
    {
      "epoch": 4.217339192968438,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 10556
    },
    {
      "epoch": 4.217738713543747,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 10557
    },
    {
      "epoch": 4.218138234119057,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 10558
    },
    {
      "epoch": 4.218537754694367,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 10559
    },
    {
      "epoch": 4.218937275269677,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 10560
    },
    {
      "epoch": 4.219336795844986,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 10561
    },
    {
      "epoch": 4.2197363164202955,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 10562
    },
    {
      "epoch": 4.220135836995605,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 10563
    },
    {
      "epoch": 4.220535357570915,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 10564
    },
    {
      "epoch": 4.220934878146225,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 10565
    },
    {
      "epoch": 4.221334398721535,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 10566
    },
    {
      "epoch": 4.2217339192968435,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 10567
    },
    {
      "epoch": 4.222133439872153,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 10568
    },
    {
      "epoch": 4.222532960447463,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 10569
    },
    {
      "epoch": 4.222932481022773,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 10570
    },
    {
      "epoch": 4.223332001598083,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 10571
    },
    {
      "epoch": 4.2237315221733915,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 10572
    },
    {
      "epoch": 4.224131042748701,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 10573
    },
    {
      "epoch": 4.224530563324011,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 10574
    },
    {
      "epoch": 4.224930083899321,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 10575
    },
    {
      "epoch": 4.225329604474631,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 10576
    },
    {
      "epoch": 4.2257291250499405,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 10577
    },
    {
      "epoch": 4.226128645625249,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0724,
      "step": 10578
    },
    {
      "epoch": 4.226528166200559,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 10579
    },
    {
      "epoch": 4.226927686775869,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 10580
    },
    {
      "epoch": 4.227327207351179,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 10581
    },
    {
      "epoch": 4.2277267279264885,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 10582
    },
    {
      "epoch": 4.228126248501797,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 10583
    },
    {
      "epoch": 4.228525769077107,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 10584
    },
    {
      "epoch": 4.228925289652417,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 10585
    },
    {
      "epoch": 4.229324810227727,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 10586
    },
    {
      "epoch": 4.229724330803037,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 10587
    },
    {
      "epoch": 4.230123851378346,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 10588
    },
    {
      "epoch": 4.230523371953655,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 10589
    },
    {
      "epoch": 4.230922892528965,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 10590
    },
    {
      "epoch": 4.231322413104275,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 10591
    },
    {
      "epoch": 4.231721933679585,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 10592
    },
    {
      "epoch": 4.232121454254894,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 10593
    },
    {
      "epoch": 4.232520974830204,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 10594
    },
    {
      "epoch": 4.232920495405513,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 10595
    },
    {
      "epoch": 4.233320015980823,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 10596
    },
    {
      "epoch": 4.233719536556133,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 10597
    },
    {
      "epoch": 4.234119057131442,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 10598
    },
    {
      "epoch": 4.234518577706752,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 10599
    },
    {
      "epoch": 4.234918098282061,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 10600
    },
    {
      "epoch": 4.235317618857371,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 10601
    },
    {
      "epoch": 4.235717139432681,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 10602
    },
    {
      "epoch": 4.2361166600079905,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 10603
    },
    {
      "epoch": 4.2365161805833,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 10604
    },
    {
      "epoch": 4.23691570115861,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 10605
    },
    {
      "epoch": 4.237315221733919,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 10606
    },
    {
      "epoch": 4.237714742309229,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 10607
    },
    {
      "epoch": 4.2381142628845385,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 10608
    },
    {
      "epoch": 4.238513783459848,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 10609
    },
    {
      "epoch": 4.238913304035158,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 10610
    },
    {
      "epoch": 4.239312824610467,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 10611
    },
    {
      "epoch": 4.239712345185777,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 10612
    },
    {
      "epoch": 4.240111865761087,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 10613
    },
    {
      "epoch": 4.240511386336396,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 10614
    },
    {
      "epoch": 4.240910906911706,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 10615
    },
    {
      "epoch": 4.241310427487016,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 10616
    },
    {
      "epoch": 4.241709948062325,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0667,
      "step": 10617
    },
    {
      "epoch": 4.242109468637635,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 10618
    },
    {
      "epoch": 4.242508989212944,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 10619
    },
    {
      "epoch": 4.242908509788254,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 10620
    },
    {
      "epoch": 4.243308030363564,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 10621
    },
    {
      "epoch": 4.243707550938874,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 10622
    },
    {
      "epoch": 4.244107071514183,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.06,
      "step": 10623
    },
    {
      "epoch": 4.244506592089492,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 10624
    },
    {
      "epoch": 4.244906112664802,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 10625
    },
    {
      "epoch": 4.245305633240112,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 10626
    },
    {
      "epoch": 4.245705153815422,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 10627
    },
    {
      "epoch": 4.246104674390731,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 10628
    },
    {
      "epoch": 4.2465041949660405,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 10629
    },
    {
      "epoch": 4.24690371554135,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 10630
    },
    {
      "epoch": 4.24730323611666,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 10631
    },
    {
      "epoch": 4.24770275669197,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 10632
    },
    {
      "epoch": 4.24810227726728,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 10633
    },
    {
      "epoch": 4.2485017978425885,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 10634
    },
    {
      "epoch": 4.248901318417898,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 10635
    },
    {
      "epoch": 4.249300838993208,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 10636
    },
    {
      "epoch": 4.249700359568518,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 10637
    },
    {
      "epoch": 4.250099880143828,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 10638
    },
    {
      "epoch": 4.2504994007191375,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 10639
    },
    {
      "epoch": 4.250898921294446,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 10640
    },
    {
      "epoch": 4.251298441869756,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 10641
    },
    {
      "epoch": 4.251697962445066,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 10642
    },
    {
      "epoch": 4.252097483020376,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 10643
    },
    {
      "epoch": 4.2524970035956855,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 10644
    },
    {
      "epoch": 4.252896524170994,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 10645
    },
    {
      "epoch": 4.253296044746304,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 10646
    },
    {
      "epoch": 4.253695565321614,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 10647
    },
    {
      "epoch": 4.254095085896924,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 10648
    },
    {
      "epoch": 4.2544946064722335,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 10649
    },
    {
      "epoch": 4.254894127047543,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 10650
    },
    {
      "epoch": 4.255293647622852,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 10651
    },
    {
      "epoch": 4.255693168198162,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 10652
    },
    {
      "epoch": 4.256092688773472,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 10653
    },
    {
      "epoch": 4.256492209348782,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 10654
    },
    {
      "epoch": 4.256891729924091,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 10655
    },
    {
      "epoch": 4.257291250499401,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 10656
    },
    {
      "epoch": 4.25769077107471,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 10657
    },
    {
      "epoch": 4.25809029165002,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 10658
    },
    {
      "epoch": 4.25848981222533,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 10659
    },
    {
      "epoch": 4.258889332800639,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 10660
    },
    {
      "epoch": 4.259288853375949,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 10661
    },
    {
      "epoch": 4.259688373951258,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 10662
    },
    {
      "epoch": 4.260087894526568,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 10663
    },
    {
      "epoch": 4.260487415101878,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 10664
    },
    {
      "epoch": 4.2608869356771875,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0646,
      "step": 10665
    },
    {
      "epoch": 4.261286456252497,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 10666
    },
    {
      "epoch": 4.261685976827807,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0626,
      "step": 10667
    },
    {
      "epoch": 4.262085497403116,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 10668
    },
    {
      "epoch": 4.262485017978426,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 10669
    },
    {
      "epoch": 4.2628845385537355,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 10670
    },
    {
      "epoch": 4.263284059129045,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 10671
    },
    {
      "epoch": 4.263683579704355,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 10672
    },
    {
      "epoch": 4.264083100279665,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.063,
      "step": 10673
    },
    {
      "epoch": 4.264482620854974,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 10674
    },
    {
      "epoch": 4.2648821414302835,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 10675
    },
    {
      "epoch": 4.265281662005593,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 10676
    },
    {
      "epoch": 4.265681182580903,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 10677
    },
    {
      "epoch": 4.266080703156213,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 10678
    },
    {
      "epoch": 4.266480223731522,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 10679
    },
    {
      "epoch": 4.266879744306832,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 10680
    },
    {
      "epoch": 4.267279264882141,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 10681
    },
    {
      "epoch": 4.267678785457451,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 10682
    },
    {
      "epoch": 4.268078306032761,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 10683
    },
    {
      "epoch": 4.26847782660807,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 10684
    },
    {
      "epoch": 4.26887734718338,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 10685
    },
    {
      "epoch": 4.269276867758689,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 10686
    },
    {
      "epoch": 4.269676388333999,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 10687
    },
    {
      "epoch": 4.270075908909309,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 10688
    },
    {
      "epoch": 4.270475429484619,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 10689
    },
    {
      "epoch": 4.270874950059928,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 10690
    },
    {
      "epoch": 4.2712744706352375,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 10691
    },
    {
      "epoch": 4.271673991210547,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 10692
    },
    {
      "epoch": 4.272073511785857,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 10693
    },
    {
      "epoch": 4.272473032361167,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 10694
    },
    {
      "epoch": 4.272872552936477,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 10695
    },
    {
      "epoch": 4.2732720735117855,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 10696
    },
    {
      "epoch": 4.273671594087095,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 10697
    },
    {
      "epoch": 4.274071114662405,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 10698
    },
    {
      "epoch": 4.274470635237715,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 10699
    },
    {
      "epoch": 4.274870155813025,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 10700
    },
    {
      "epoch": 4.2752696763883336,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 10701
    },
    {
      "epoch": 4.275669196963643,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 10702
    },
    {
      "epoch": 4.276068717538953,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 10703
    },
    {
      "epoch": 4.276468238114263,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 10704
    },
    {
      "epoch": 4.276867758689573,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 10705
    },
    {
      "epoch": 4.2772672792648825,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 10706
    },
    {
      "epoch": 4.277666799840191,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 10707
    },
    {
      "epoch": 4.278066320415501,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 10708
    },
    {
      "epoch": 4.278465840990811,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 10709
    },
    {
      "epoch": 4.278865361566121,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 10710
    },
    {
      "epoch": 4.2792648821414305,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 10711
    },
    {
      "epoch": 4.27966440271674,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 10712
    },
    {
      "epoch": 4.280063923292049,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 10713
    },
    {
      "epoch": 4.280463443867359,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 10714
    },
    {
      "epoch": 4.280862964442669,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 10715
    },
    {
      "epoch": 4.281262485017979,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 10716
    },
    {
      "epoch": 4.281662005593288,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 10717
    },
    {
      "epoch": 4.282061526168597,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 10718
    },
    {
      "epoch": 4.282461046743907,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 10719
    },
    {
      "epoch": 4.282860567319217,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 10720
    },
    {
      "epoch": 4.283260087894527,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 10721
    },
    {
      "epoch": 4.283659608469836,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 10722
    },
    {
      "epoch": 4.284059129045146,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 10723
    },
    {
      "epoch": 4.284458649620455,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 10724
    },
    {
      "epoch": 4.284858170195765,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 10725
    },
    {
      "epoch": 4.285257690771075,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 10726
    },
    {
      "epoch": 4.285657211346384,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 10727
    },
    {
      "epoch": 4.286056731921694,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 10728
    },
    {
      "epoch": 4.286456252497004,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 10729
    },
    {
      "epoch": 4.286855773072313,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 10730
    },
    {
      "epoch": 4.287255293647623,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 10731
    },
    {
      "epoch": 4.2876548142229325,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 10732
    },
    {
      "epoch": 4.288054334798242,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 10733
    },
    {
      "epoch": 4.288453855373552,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 10734
    },
    {
      "epoch": 4.288853375948861,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 10735
    },
    {
      "epoch": 4.289252896524171,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 10736
    },
    {
      "epoch": 4.2896524170994805,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 10737
    },
    {
      "epoch": 4.29005193767479,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 10738
    },
    {
      "epoch": 4.2904514582501,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 10739
    },
    {
      "epoch": 4.29085097882541,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 10740
    },
    {
      "epoch": 4.291250499400719,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 10741
    },
    {
      "epoch": 4.291650019976029,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 10742
    },
    {
      "epoch": 4.292049540551338,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 10743
    },
    {
      "epoch": 4.292449061126648,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 10744
    },
    {
      "epoch": 4.292848581701958,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 10745
    },
    {
      "epoch": 4.293248102277268,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 10746
    },
    {
      "epoch": 4.293647622852577,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 10747
    },
    {
      "epoch": 4.294047143427886,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 10748
    },
    {
      "epoch": 4.294446664003196,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 10749
    },
    {
      "epoch": 4.294846184578506,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 10750
    },
    {
      "epoch": 4.295245705153816,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 10751
    },
    {
      "epoch": 4.295645225729125,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 10752
    },
    {
      "epoch": 4.2960447463044344,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 10753
    },
    {
      "epoch": 4.296444266879744,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 10754
    },
    {
      "epoch": 4.296843787455054,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 10755
    },
    {
      "epoch": 4.297243308030364,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 10756
    },
    {
      "epoch": 4.297642828605674,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 10757
    },
    {
      "epoch": 4.2980423491809825,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 10758
    },
    {
      "epoch": 4.298441869756292,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 10759
    },
    {
      "epoch": 4.298841390331602,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 10760
    },
    {
      "epoch": 4.299240910906912,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 10761
    },
    {
      "epoch": 4.299640431482222,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 10762
    },
    {
      "epoch": 4.300039952057531,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 10763
    },
    {
      "epoch": 4.30043947263284,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 10764
    },
    {
      "epoch": 4.30083899320815,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 10765
    },
    {
      "epoch": 4.30123851378346,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 10766
    },
    {
      "epoch": 4.30163803435877,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 10767
    },
    {
      "epoch": 4.3020375549340795,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 10768
    },
    {
      "epoch": 4.302437075509388,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 10769
    },
    {
      "epoch": 4.302836596084698,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 10770
    },
    {
      "epoch": 4.303236116660008,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 10771
    },
    {
      "epoch": 4.303635637235318,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 10772
    },
    {
      "epoch": 4.3040351578106275,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 10773
    },
    {
      "epoch": 4.304434678385936,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 10774
    },
    {
      "epoch": 4.304834198961246,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 10775
    },
    {
      "epoch": 4.305233719536556,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 10776
    },
    {
      "epoch": 4.305633240111866,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 10777
    },
    {
      "epoch": 4.3060327606871756,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 10778
    },
    {
      "epoch": 4.306432281262485,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 10779
    },
    {
      "epoch": 4.306831801837794,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 10780
    },
    {
      "epoch": 4.307231322413104,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 10781
    },
    {
      "epoch": 4.307630842988414,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 10782
    },
    {
      "epoch": 4.308030363563724,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 10783
    },
    {
      "epoch": 4.308429884139033,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 10784
    },
    {
      "epoch": 4.308829404714343,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 10785
    },
    {
      "epoch": 4.309228925289652,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 10786
    },
    {
      "epoch": 4.309628445864962,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 10787
    },
    {
      "epoch": 4.310027966440272,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 10788
    },
    {
      "epoch": 4.310427487015581,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 10789
    },
    {
      "epoch": 4.310827007590891,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0682,
      "step": 10790
    },
    {
      "epoch": 4.3112265281662,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 10791
    },
    {
      "epoch": 4.31162604874151,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 10792
    },
    {
      "epoch": 4.31202556931682,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 10793
    },
    {
      "epoch": 4.3124250898921295,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 10794
    },
    {
      "epoch": 4.312824610467439,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 10795
    },
    {
      "epoch": 4.313224131042749,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 10796
    },
    {
      "epoch": 4.313623651618058,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 10797
    },
    {
      "epoch": 4.314023172193368,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 10798
    },
    {
      "epoch": 4.3144226927686775,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 10799
    },
    {
      "epoch": 4.314822213343987,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 10800
    },
    {
      "epoch": 4.315221733919297,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 10801
    },
    {
      "epoch": 4.315621254494607,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 10802
    },
    {
      "epoch": 4.316020775069916,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 10803
    },
    {
      "epoch": 4.316420295645226,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 10804
    },
    {
      "epoch": 4.316819816220535,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 10805
    },
    {
      "epoch": 4.317219336795845,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 10806
    },
    {
      "epoch": 4.317618857371155,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 10807
    },
    {
      "epoch": 4.318018377946464,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 10808
    },
    {
      "epoch": 4.318417898521774,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 10809
    },
    {
      "epoch": 4.318817419097083,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 10810
    },
    {
      "epoch": 4.319216939672393,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 10811
    },
    {
      "epoch": 4.319616460247703,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 10812
    },
    {
      "epoch": 4.320015980823013,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 10813
    },
    {
      "epoch": 4.320415501398322,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 10814
    },
    {
      "epoch": 4.320815021973631,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 10815
    },
    {
      "epoch": 4.321214542548941,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 10816
    },
    {
      "epoch": 4.321614063124251,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 10817
    },
    {
      "epoch": 4.322013583699561,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 10818
    },
    {
      "epoch": 4.322413104274871,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 10819
    },
    {
      "epoch": 4.3228126248501795,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 10820
    },
    {
      "epoch": 4.323212145425489,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 10821
    },
    {
      "epoch": 4.323611666000799,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 10822
    },
    {
      "epoch": 4.324011186576109,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 10823
    },
    {
      "epoch": 4.324410707151419,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 10824
    },
    {
      "epoch": 4.3248102277267275,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 10825
    },
    {
      "epoch": 4.325209748302037,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 10826
    },
    {
      "epoch": 4.325609268877347,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 10827
    },
    {
      "epoch": 4.326008789452657,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 10828
    },
    {
      "epoch": 4.326408310027967,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 10829
    },
    {
      "epoch": 4.3268078306032765,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 10830
    },
    {
      "epoch": 4.327207351178585,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 10831
    },
    {
      "epoch": 4.327606871753895,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 10832
    },
    {
      "epoch": 4.328006392329205,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 10833
    },
    {
      "epoch": 4.328405912904515,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 10834
    },
    {
      "epoch": 4.3288054334798245,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 10835
    },
    {
      "epoch": 4.329204954055134,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 10836
    },
    {
      "epoch": 4.329604474630443,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 10837
    },
    {
      "epoch": 4.330003995205753,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 10838
    },
    {
      "epoch": 4.330403515781063,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 10839
    },
    {
      "epoch": 4.3308030363563725,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 10840
    },
    {
      "epoch": 4.331202556931682,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 10841
    },
    {
      "epoch": 4.331602077506991,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 10842
    },
    {
      "epoch": 4.332001598082301,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 10843
    },
    {
      "epoch": 4.332401118657611,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 10844
    },
    {
      "epoch": 4.332800639232921,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 10845
    },
    {
      "epoch": 4.33320015980823,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 10846
    },
    {
      "epoch": 4.33359968038354,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 10847
    },
    {
      "epoch": 4.333999200958849,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 10848
    },
    {
      "epoch": 4.334398721534159,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 10849
    },
    {
      "epoch": 4.334798242109469,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 10850
    },
    {
      "epoch": 4.335197762684778,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 10851
    },
    {
      "epoch": 4.335597283260088,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 10852
    },
    {
      "epoch": 4.335996803835398,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 10853
    },
    {
      "epoch": 4.336396324410707,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 10854
    },
    {
      "epoch": 4.336795844986017,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 10855
    },
    {
      "epoch": 4.3371953655613265,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 10856
    },
    {
      "epoch": 4.337594886136636,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 10857
    },
    {
      "epoch": 4.337994406711946,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 10858
    },
    {
      "epoch": 4.338393927287255,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 10859
    },
    {
      "epoch": 4.338793447862565,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 10860
    },
    {
      "epoch": 4.3391929684378745,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 10861
    },
    {
      "epoch": 4.339592489013184,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 10862
    },
    {
      "epoch": 4.339992009588494,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 10863
    },
    {
      "epoch": 4.340391530163803,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 10864
    },
    {
      "epoch": 4.340791050739113,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 10865
    },
    {
      "epoch": 4.3411905713144225,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 10866
    },
    {
      "epoch": 4.341590091889732,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 10867
    },
    {
      "epoch": 4.341989612465042,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 10868
    },
    {
      "epoch": 4.342389133040352,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 10869
    },
    {
      "epoch": 4.342788653615661,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 10870
    },
    {
      "epoch": 4.343188174190971,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 10871
    },
    {
      "epoch": 4.34358769476628,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 10872
    },
    {
      "epoch": 4.34398721534159,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 10873
    },
    {
      "epoch": 4.3443867359169,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 10874
    },
    {
      "epoch": 4.34478625649221,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 10875
    },
    {
      "epoch": 4.345185777067519,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 10876
    },
    {
      "epoch": 4.345585297642828,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 10877
    },
    {
      "epoch": 4.345984818218138,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 10878
    },
    {
      "epoch": 4.346384338793448,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 10879
    },
    {
      "epoch": 4.346783859368758,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 10880
    },
    {
      "epoch": 4.347183379944067,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 10881
    },
    {
      "epoch": 4.3475829005193765,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 10882
    },
    {
      "epoch": 4.347982421094686,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 10883
    },
    {
      "epoch": 4.348381941669996,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 10884
    },
    {
      "epoch": 4.348781462245306,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 10885
    },
    {
      "epoch": 4.349180982820616,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 10886
    },
    {
      "epoch": 4.3495805033959245,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 10887
    },
    {
      "epoch": 4.349980023971234,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 10888
    },
    {
      "epoch": 4.350379544546544,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 10889
    },
    {
      "epoch": 4.350779065121854,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 10890
    },
    {
      "epoch": 4.351178585697164,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 10891
    },
    {
      "epoch": 4.351578106272473,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 10892
    },
    {
      "epoch": 4.351977626847782,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 10893
    },
    {
      "epoch": 4.352377147423092,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 10894
    },
    {
      "epoch": 4.352776667998402,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 10895
    },
    {
      "epoch": 4.353176188573712,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 10896
    },
    {
      "epoch": 4.3535757091490215,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 10897
    },
    {
      "epoch": 4.35397522972433,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 10898
    },
    {
      "epoch": 4.35437475029964,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 10899
    },
    {
      "epoch": 4.35477427087495,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 10900
    },
    {
      "epoch": 4.35517379145026,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 10901
    },
    {
      "epoch": 4.3555733120255695,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 10902
    },
    {
      "epoch": 4.355972832600879,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 10903
    },
    {
      "epoch": 4.356372353176188,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 10904
    },
    {
      "epoch": 4.356771873751498,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 10905
    },
    {
      "epoch": 4.357171394326808,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 10906
    },
    {
      "epoch": 4.357570914902118,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 10907
    },
    {
      "epoch": 4.357970435477427,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 10908
    },
    {
      "epoch": 4.358369956052737,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 10909
    },
    {
      "epoch": 4.358769476628046,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 10910
    },
    {
      "epoch": 4.359168997203356,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 10911
    },
    {
      "epoch": 4.359568517778666,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 10912
    },
    {
      "epoch": 4.359968038353975,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 10913
    },
    {
      "epoch": 4.360367558929285,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 10914
    },
    {
      "epoch": 4.360767079504594,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 10915
    },
    {
      "epoch": 4.361166600079904,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 10916
    },
    {
      "epoch": 4.361566120655214,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 10917
    },
    {
      "epoch": 4.361965641230523,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 10918
    },
    {
      "epoch": 4.362365161805833,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 10919
    },
    {
      "epoch": 4.362764682381143,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 10920
    },
    {
      "epoch": 4.363164202956452,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 10921
    },
    {
      "epoch": 4.363563723531762,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 10922
    },
    {
      "epoch": 4.3639632441070715,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 10923
    },
    {
      "epoch": 4.364362764682381,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 10924
    },
    {
      "epoch": 4.364762285257691,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 10925
    },
    {
      "epoch": 4.365161805833001,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 10926
    },
    {
      "epoch": 4.36556132640831,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 10927
    },
    {
      "epoch": 4.3659608469836195,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 10928
    },
    {
      "epoch": 4.366360367558929,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 10929
    },
    {
      "epoch": 4.366759888134239,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 10930
    },
    {
      "epoch": 4.367159408709549,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 10931
    },
    {
      "epoch": 4.367558929284858,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 10932
    },
    {
      "epoch": 4.367958449860168,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 10933
    },
    {
      "epoch": 4.368357970435477,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 10934
    },
    {
      "epoch": 4.368757491010787,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 10935
    },
    {
      "epoch": 4.369157011586097,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 10936
    },
    {
      "epoch": 4.369556532161407,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 10937
    },
    {
      "epoch": 4.369956052736716,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 10938
    },
    {
      "epoch": 4.370355573312025,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 10939
    },
    {
      "epoch": 4.370755093887335,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 10940
    },
    {
      "epoch": 4.371154614462645,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 10941
    },
    {
      "epoch": 4.371554135037955,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 10942
    },
    {
      "epoch": 4.3719536556132645,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 10943
    },
    {
      "epoch": 4.372353176188573,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 10944
    },
    {
      "epoch": 4.372752696763883,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 10945
    },
    {
      "epoch": 4.373152217339193,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 10946
    },
    {
      "epoch": 4.373551737914503,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 10947
    },
    {
      "epoch": 4.373951258489813,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 10948
    },
    {
      "epoch": 4.3743507790651215,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 10949
    },
    {
      "epoch": 4.374750299640431,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 10950
    },
    {
      "epoch": 4.375149820215741,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 10951
    },
    {
      "epoch": 4.375549340791051,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 10952
    },
    {
      "epoch": 4.375948861366361,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 10953
    },
    {
      "epoch": 4.3763483819416695,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 10954
    },
    {
      "epoch": 4.376747902516979,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 10955
    },
    {
      "epoch": 4.377147423092289,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 10956
    },
    {
      "epoch": 4.377546943667599,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 10957
    },
    {
      "epoch": 4.377946464242909,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 10958
    },
    {
      "epoch": 4.3783459848182185,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 10959
    },
    {
      "epoch": 4.378745505393527,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 10960
    },
    {
      "epoch": 4.379145025968837,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 10961
    },
    {
      "epoch": 4.379544546544147,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 10962
    },
    {
      "epoch": 4.379944067119457,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 10963
    },
    {
      "epoch": 4.3803435876947665,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 10964
    },
    {
      "epoch": 4.380743108270076,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 10965
    },
    {
      "epoch": 4.381142628845385,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 10966
    },
    {
      "epoch": 4.381542149420695,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 10967
    },
    {
      "epoch": 4.381941669996005,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 10968
    },
    {
      "epoch": 4.3823411905713145,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 10969
    },
    {
      "epoch": 4.382740711146624,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 10970
    },
    {
      "epoch": 4.383140231721933,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 10971
    },
    {
      "epoch": 4.383539752297243,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 10972
    },
    {
      "epoch": 4.383939272872553,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 10973
    },
    {
      "epoch": 4.384338793447863,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 10974
    },
    {
      "epoch": 4.384738314023172,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 10975
    },
    {
      "epoch": 4.385137834598482,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 10976
    },
    {
      "epoch": 4.385537355173791,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 10977
    },
    {
      "epoch": 4.385936875749101,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 10978
    },
    {
      "epoch": 4.386336396324411,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 10979
    },
    {
      "epoch": 4.38673591689972,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 10980
    },
    {
      "epoch": 4.38713543747503,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 10981
    },
    {
      "epoch": 4.38753495805034,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 10982
    },
    {
      "epoch": 4.387934478625649,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 10983
    },
    {
      "epoch": 4.388333999200959,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 10984
    },
    {
      "epoch": 4.3887335197762685,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 10985
    },
    {
      "epoch": 4.389133040351578,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 10986
    },
    {
      "epoch": 4.389532560926888,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.064,
      "step": 10987
    },
    {
      "epoch": 4.389932081502197,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 10988
    },
    {
      "epoch": 4.390331602077507,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 10989
    },
    {
      "epoch": 4.3907311226528165,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 10990
    },
    {
      "epoch": 4.391130643228126,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 10991
    },
    {
      "epoch": 4.391530163803436,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 10992
    },
    {
      "epoch": 4.391929684378746,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 10993
    },
    {
      "epoch": 4.392329204954055,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 10994
    },
    {
      "epoch": 4.3927287255293646,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 10995
    },
    {
      "epoch": 4.393128246104674,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 10996
    },
    {
      "epoch": 4.393527766679984,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 10997
    },
    {
      "epoch": 4.393927287255294,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 10998
    },
    {
      "epoch": 4.394326807830604,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 10999
    },
    {
      "epoch": 4.394726328405913,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 11000
    },
    {
      "epoch": 4.395125848981222,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 11001
    },
    {
      "epoch": 4.395525369556532,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0627,
      "step": 11002
    },
    {
      "epoch": 4.395924890131842,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 11003
    },
    {
      "epoch": 4.396324410707152,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 11004
    },
    {
      "epoch": 4.396723931282461,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 11005
    },
    {
      "epoch": 4.39712345185777,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 11006
    },
    {
      "epoch": 4.39752297243308,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 11007
    },
    {
      "epoch": 4.39792249300839,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 11008
    },
    {
      "epoch": 4.3983220135837,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 11009
    },
    {
      "epoch": 4.39872153415901,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 11010
    },
    {
      "epoch": 4.3991210547343185,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 11011
    },
    {
      "epoch": 4.399520575309628,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 11012
    },
    {
      "epoch": 4.399920095884938,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 11013
    },
    {
      "epoch": 4.400319616460248,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 11014
    },
    {
      "epoch": 4.400719137035558,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 11015
    },
    {
      "epoch": 4.401118657610867,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 11016
    },
    {
      "epoch": 4.401518178186176,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 11017
    },
    {
      "epoch": 4.401917698761486,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0642,
      "step": 11018
    },
    {
      "epoch": 4.402317219336796,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 11019
    },
    {
      "epoch": 4.402716739912106,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 11020
    },
    {
      "epoch": 4.4031162604874154,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 11021
    },
    {
      "epoch": 4.403515781062724,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 11022
    },
    {
      "epoch": 4.403915301638034,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 11023
    },
    {
      "epoch": 4.404314822213344,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 11024
    },
    {
      "epoch": 4.404714342788654,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 11025
    },
    {
      "epoch": 4.4051138633639635,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 11026
    },
    {
      "epoch": 4.405513383939273,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 11027
    },
    {
      "epoch": 4.405912904514582,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 11028
    },
    {
      "epoch": 4.406312425089892,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 11029
    },
    {
      "epoch": 4.406711945665202,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 11030
    },
    {
      "epoch": 4.4071114662405115,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 11031
    },
    {
      "epoch": 4.407510986815821,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 11032
    },
    {
      "epoch": 4.407910507391131,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 11033
    },
    {
      "epoch": 4.40831002796644,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 11034
    },
    {
      "epoch": 4.40870954854175,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 11035
    },
    {
      "epoch": 4.40910906911706,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 11036
    },
    {
      "epoch": 4.409508589692369,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 11037
    },
    {
      "epoch": 4.409908110267679,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 11038
    },
    {
      "epoch": 4.410307630842988,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 11039
    },
    {
      "epoch": 4.410707151418298,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 11040
    },
    {
      "epoch": 4.411106671993608,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 11041
    },
    {
      "epoch": 4.411506192568917,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 11042
    },
    {
      "epoch": 4.411905713144227,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 11043
    },
    {
      "epoch": 4.412305233719536,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 11044
    },
    {
      "epoch": 4.412704754294846,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 11045
    },
    {
      "epoch": 4.413104274870156,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 11046
    },
    {
      "epoch": 4.4135037954454654,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 11047
    },
    {
      "epoch": 4.413903316020775,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 11048
    },
    {
      "epoch": 4.414302836596085,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 11049
    },
    {
      "epoch": 4.414702357171394,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 11050
    },
    {
      "epoch": 4.415101877746704,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 11051
    },
    {
      "epoch": 4.4155013983220135,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 11052
    },
    {
      "epoch": 4.415900918897323,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 11053
    },
    {
      "epoch": 4.416300439472633,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 11054
    },
    {
      "epoch": 4.416699960047943,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 11055
    },
    {
      "epoch": 4.417099480623252,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 11056
    },
    {
      "epoch": 4.4174990011985615,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 11057
    },
    {
      "epoch": 4.417898521773871,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 11058
    },
    {
      "epoch": 4.418298042349181,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 11059
    },
    {
      "epoch": 4.418697562924491,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 11060
    },
    {
      "epoch": 4.4190970834998,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 11061
    },
    {
      "epoch": 4.41949660407511,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 11062
    },
    {
      "epoch": 4.419896124650419,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 11063
    },
    {
      "epoch": 4.420295645225729,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 11064
    },
    {
      "epoch": 4.420695165801039,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 11065
    },
    {
      "epoch": 4.421094686376349,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 11066
    },
    {
      "epoch": 4.421494206951658,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 11067
    },
    {
      "epoch": 4.421893727526967,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 11068
    },
    {
      "epoch": 4.422293248102277,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 11069
    },
    {
      "epoch": 4.422692768677587,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 11070
    },
    {
      "epoch": 4.423092289252897,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 11071
    },
    {
      "epoch": 4.4234918098282066,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 11072
    },
    {
      "epoch": 4.4238913304035155,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 11073
    },
    {
      "epoch": 4.424290850978825,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 11074
    },
    {
      "epoch": 4.424690371554135,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 11075
    },
    {
      "epoch": 4.425089892129445,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 11076
    },
    {
      "epoch": 4.425489412704755,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 11077
    },
    {
      "epoch": 4.4258889332800635,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 11078
    },
    {
      "epoch": 4.426288453855373,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 11079
    },
    {
      "epoch": 4.426687974430683,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 11080
    },
    {
      "epoch": 4.427087495005993,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 11081
    },
    {
      "epoch": 4.427487015581303,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 11082
    },
    {
      "epoch": 4.427886536156612,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 11083
    },
    {
      "epoch": 4.428286056731921,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 11084
    },
    {
      "epoch": 4.428685577307231,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 11085
    },
    {
      "epoch": 4.429085097882541,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 11086
    },
    {
      "epoch": 4.429484618457851,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 11087
    },
    {
      "epoch": 4.4298841390331605,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 11088
    },
    {
      "epoch": 4.43028365960847,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 11089
    },
    {
      "epoch": 4.430683180183779,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 11090
    },
    {
      "epoch": 4.431082700759089,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 11091
    },
    {
      "epoch": 4.431482221334399,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 11092
    },
    {
      "epoch": 4.4318817419097085,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 11093
    },
    {
      "epoch": 4.432281262485018,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 11094
    },
    {
      "epoch": 4.432680783060327,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 11095
    },
    {
      "epoch": 4.433080303635637,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 11096
    },
    {
      "epoch": 4.433479824210947,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 11097
    },
    {
      "epoch": 4.433879344786257,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 11098
    },
    {
      "epoch": 4.434278865361566,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 11099
    },
    {
      "epoch": 4.434678385936876,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 11100
    },
    {
      "epoch": 4.435077906512185,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 11101
    },
    {
      "epoch": 4.435477427087495,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 11102
    },
    {
      "epoch": 4.435876947662805,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 11103
    },
    {
      "epoch": 4.436276468238114,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 11104
    },
    {
      "epoch": 4.436675988813424,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 11105
    },
    {
      "epoch": 4.437075509388734,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 11106
    },
    {
      "epoch": 4.437475029964043,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 11107
    },
    {
      "epoch": 4.437874550539353,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 11108
    },
    {
      "epoch": 4.438274071114662,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 11109
    },
    {
      "epoch": 4.438673591689972,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 11110
    },
    {
      "epoch": 4.439073112265282,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 11111
    },
    {
      "epoch": 4.439472632840591,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 11112
    },
    {
      "epoch": 4.439872153415901,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 11113
    },
    {
      "epoch": 4.4402716739912105,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 11114
    },
    {
      "epoch": 4.44067119456652,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 11115
    },
    {
      "epoch": 4.44107071514183,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 11116
    },
    {
      "epoch": 4.44147023571714,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 11117
    },
    {
      "epoch": 4.441869756292449,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 11118
    },
    {
      "epoch": 4.4422692768677585,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 11119
    },
    {
      "epoch": 4.442668797443068,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 11120
    },
    {
      "epoch": 4.443068318018378,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 11121
    },
    {
      "epoch": 4.443467838593688,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 11122
    },
    {
      "epoch": 4.443867359168998,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 11123
    },
    {
      "epoch": 4.444266879744307,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 11124
    },
    {
      "epoch": 4.444666400319616,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 11125
    },
    {
      "epoch": 4.445065920894926,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 11126
    },
    {
      "epoch": 4.445465441470236,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 11127
    },
    {
      "epoch": 4.445864962045546,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 11128
    },
    {
      "epoch": 4.446264482620855,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 11129
    },
    {
      "epoch": 4.446664003196164,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 11130
    },
    {
      "epoch": 4.447063523771474,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 11131
    },
    {
      "epoch": 4.447463044346784,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 11132
    },
    {
      "epoch": 4.447862564922094,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 11133
    },
    {
      "epoch": 4.448262085497403,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 11134
    },
    {
      "epoch": 4.448661606072712,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 11135
    },
    {
      "epoch": 4.449061126648022,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 11136
    },
    {
      "epoch": 4.449460647223332,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 11137
    },
    {
      "epoch": 4.449860167798642,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 11138
    },
    {
      "epoch": 4.450259688373952,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 11139
    },
    {
      "epoch": 4.4506592089492605,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 11140
    },
    {
      "epoch": 4.45105872952457,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 11141
    },
    {
      "epoch": 4.45145825009988,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 11142
    },
    {
      "epoch": 4.45185777067519,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 11143
    },
    {
      "epoch": 4.4522572912505,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 11144
    },
    {
      "epoch": 4.452656811825809,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 11145
    },
    {
      "epoch": 4.453056332401118,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 11146
    },
    {
      "epoch": 4.453455852976428,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 11147
    },
    {
      "epoch": 4.453855373551738,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 11148
    },
    {
      "epoch": 4.454254894127048,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 11149
    },
    {
      "epoch": 4.4546544147023575,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 11150
    },
    {
      "epoch": 4.455053935277666,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 11151
    },
    {
      "epoch": 4.455453455852976,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 11152
    },
    {
      "epoch": 4.455852976428286,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 11153
    },
    {
      "epoch": 4.456252497003596,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 11154
    },
    {
      "epoch": 4.4566520175789055,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 11155
    },
    {
      "epoch": 4.457051538154215,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 11156
    },
    {
      "epoch": 4.457451058729524,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 11157
    },
    {
      "epoch": 4.457850579304834,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 11158
    },
    {
      "epoch": 4.458250099880144,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 11159
    },
    {
      "epoch": 4.4586496204554535,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 11160
    },
    {
      "epoch": 4.459049141030763,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 11161
    },
    {
      "epoch": 4.459448661606073,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 11162
    },
    {
      "epoch": 4.459848182181382,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 11163
    },
    {
      "epoch": 4.460247702756692,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 11164
    },
    {
      "epoch": 4.460647223332002,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 11165
    },
    {
      "epoch": 4.461046743907311,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 11166
    },
    {
      "epoch": 4.461446264482621,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 11167
    },
    {
      "epoch": 4.46184578505793,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 11168
    },
    {
      "epoch": 4.46224530563324,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 11169
    },
    {
      "epoch": 4.46264482620855,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 11170
    },
    {
      "epoch": 4.463044346783859,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 11171
    },
    {
      "epoch": 4.463443867359169,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 11172
    },
    {
      "epoch": 4.463843387934479,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 11173
    },
    {
      "epoch": 4.464242908509788,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 11174
    },
    {
      "epoch": 4.464642429085098,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 11175
    },
    {
      "epoch": 4.4650419496604075,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 11176
    },
    {
      "epoch": 4.465441470235717,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 11177
    },
    {
      "epoch": 4.465840990811027,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 11178
    },
    {
      "epoch": 4.466240511386337,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 11179
    },
    {
      "epoch": 4.466640031961646,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 11180
    },
    {
      "epoch": 4.4670395525369555,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 11181
    },
    {
      "epoch": 4.467439073112265,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 11182
    },
    {
      "epoch": 4.467838593687575,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 11183
    },
    {
      "epoch": 4.468238114262885,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 11184
    },
    {
      "epoch": 4.468637634838194,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 11185
    },
    {
      "epoch": 4.4690371554135035,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 11186
    },
    {
      "epoch": 4.469436675988813,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 11187
    },
    {
      "epoch": 4.469836196564123,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 11188
    },
    {
      "epoch": 4.470235717139433,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 11189
    },
    {
      "epoch": 4.470635237714743,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 11190
    },
    {
      "epoch": 4.471034758290052,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 11191
    },
    {
      "epoch": 4.471434278865361,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 11192
    },
    {
      "epoch": 4.471833799440671,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 11193
    },
    {
      "epoch": 4.472233320015981,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 11194
    },
    {
      "epoch": 4.472632840591291,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 11195
    },
    {
      "epoch": 4.4730323611666005,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 11196
    },
    {
      "epoch": 4.473431881741909,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 11197
    },
    {
      "epoch": 4.473831402317219,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 11198
    },
    {
      "epoch": 4.474230922892529,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 11199
    },
    {
      "epoch": 4.474630443467839,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 11200
    },
    {
      "epoch": 4.475029964043149,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 11201
    },
    {
      "epoch": 4.4754294846184575,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 11202
    },
    {
      "epoch": 4.475829005193767,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 11203
    },
    {
      "epoch": 4.476228525769077,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 11204
    },
    {
      "epoch": 4.476628046344387,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 11205
    },
    {
      "epoch": 4.477027566919697,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 11206
    },
    {
      "epoch": 4.477427087495006,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 11207
    },
    {
      "epoch": 4.477826608070315,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 11208
    },
    {
      "epoch": 4.478226128645625,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 11209
    },
    {
      "epoch": 4.478625649220935,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 11210
    },
    {
      "epoch": 4.479025169796245,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 11211
    },
    {
      "epoch": 4.479424690371554,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 11212
    },
    {
      "epoch": 4.479824210946864,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 11213
    },
    {
      "epoch": 4.480223731522173,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 11214
    },
    {
      "epoch": 4.480623252097483,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 11215
    },
    {
      "epoch": 4.481022772672793,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 11216
    },
    {
      "epoch": 4.4814222932481025,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 11217
    },
    {
      "epoch": 4.481821813823412,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 11218
    },
    {
      "epoch": 4.482221334398721,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.065,
      "step": 11219
    },
    {
      "epoch": 4.482620854974031,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 11220
    },
    {
      "epoch": 4.483020375549341,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 11221
    },
    {
      "epoch": 4.4834198961246505,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 11222
    },
    {
      "epoch": 4.48381941669996,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 11223
    },
    {
      "epoch": 4.484218937275269,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 11224
    },
    {
      "epoch": 4.484618457850579,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 11225
    },
    {
      "epoch": 4.485017978425889,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 11226
    },
    {
      "epoch": 4.485417499001199,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 11227
    },
    {
      "epoch": 4.485817019576508,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 11228
    },
    {
      "epoch": 4.486216540151818,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 11229
    },
    {
      "epoch": 4.486616060727127,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 11230
    },
    {
      "epoch": 4.487015581302437,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 11231
    },
    {
      "epoch": 4.487415101877747,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 11232
    },
    {
      "epoch": 4.487814622453056,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 11233
    },
    {
      "epoch": 4.488214143028366,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 11234
    },
    {
      "epoch": 4.488613663603676,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 11235
    },
    {
      "epoch": 4.489013184178985,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 11236
    },
    {
      "epoch": 4.489412704754295,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 11237
    },
    {
      "epoch": 4.4898122253296044,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 11238
    },
    {
      "epoch": 4.490211745904914,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 11239
    },
    {
      "epoch": 4.490611266480224,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 11240
    },
    {
      "epoch": 4.491010787055533,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 11241
    },
    {
      "epoch": 4.491410307630843,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 11242
    },
    {
      "epoch": 4.4918098282061525,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 11243
    },
    {
      "epoch": 4.492209348781462,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 11244
    },
    {
      "epoch": 4.492608869356772,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 11245
    },
    {
      "epoch": 4.493008389932082,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 11246
    },
    {
      "epoch": 4.493407910507391,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 11247
    },
    {
      "epoch": 4.4938074310827005,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 11248
    },
    {
      "epoch": 4.49420695165801,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 11249
    },
    {
      "epoch": 4.49460647223332,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 11250
    },
    {
      "epoch": 4.49500599280863,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 11251
    },
    {
      "epoch": 4.49540551338394,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 11252
    },
    {
      "epoch": 4.495805033959249,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 11253
    },
    {
      "epoch": 4.496204554534558,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 11254
    },
    {
      "epoch": 4.496604075109868,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 11255
    },
    {
      "epoch": 4.497003595685178,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 11256
    },
    {
      "epoch": 4.497403116260488,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 11257
    },
    {
      "epoch": 4.497802636835797,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 11258
    },
    {
      "epoch": 4.498202157411106,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 11259
    },
    {
      "epoch": 4.498601677986416,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 11260
    },
    {
      "epoch": 4.499001198561726,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 11261
    },
    {
      "epoch": 4.499400719137036,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 11262
    },
    {
      "epoch": 4.4998002397123456,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 11263
    },
    {
      "epoch": 4.5001997602876544,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 11264
    },
    {
      "epoch": 4.500599280862964,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 11265
    },
    {
      "epoch": 4.500998801438274,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 11266
    },
    {
      "epoch": 4.501398322013584,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 11267
    },
    {
      "epoch": 4.501797842588894,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 11268
    },
    {
      "epoch": 4.502197363164203,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 11269
    },
    {
      "epoch": 4.502596883739512,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 11270
    },
    {
      "epoch": 4.502996404314822,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 11271
    },
    {
      "epoch": 4.503395924890132,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 11272
    },
    {
      "epoch": 4.503795445465442,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 11273
    },
    {
      "epoch": 4.504194966040751,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 11274
    },
    {
      "epoch": 4.50459448661606,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 11275
    },
    {
      "epoch": 4.50499400719137,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 11276
    },
    {
      "epoch": 4.50539352776668,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 11277
    },
    {
      "epoch": 4.50579304834199,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 11278
    },
    {
      "epoch": 4.5061925689172995,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 11279
    },
    {
      "epoch": 4.506592089492609,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 11280
    },
    {
      "epoch": 4.506991610067918,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 11281
    },
    {
      "epoch": 4.507391130643228,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 11282
    },
    {
      "epoch": 4.507790651218538,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 11283
    },
    {
      "epoch": 4.5081901717938475,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 11284
    },
    {
      "epoch": 4.508589692369157,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 11285
    },
    {
      "epoch": 4.508989212944467,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 11286
    },
    {
      "epoch": 4.509388733519776,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 11287
    },
    {
      "epoch": 4.509788254095086,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 11288
    },
    {
      "epoch": 4.5101877746703956,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 11289
    },
    {
      "epoch": 4.510587295245705,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 11290
    },
    {
      "epoch": 4.510986815821015,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 11291
    },
    {
      "epoch": 4.511386336396324,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 11292
    },
    {
      "epoch": 4.511785856971634,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 11293
    },
    {
      "epoch": 4.512185377546944,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 11294
    },
    {
      "epoch": 4.512584898122253,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 11295
    },
    {
      "epoch": 4.512984418697563,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 11296
    },
    {
      "epoch": 4.513383939272872,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 11297
    },
    {
      "epoch": 4.513783459848182,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 11298
    },
    {
      "epoch": 4.514182980423492,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 11299
    },
    {
      "epoch": 4.514582500998801,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 11300
    },
    {
      "epoch": 4.514982021574111,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 11301
    },
    {
      "epoch": 4.515381542149421,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 11302
    },
    {
      "epoch": 4.515781062724731,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 11303
    },
    {
      "epoch": 4.51618058330004,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 11304
    },
    {
      "epoch": 4.5165801038753495,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 11305
    },
    {
      "epoch": 4.516979624450659,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 11306
    },
    {
      "epoch": 4.517379145025969,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 11307
    },
    {
      "epoch": 4.517778665601279,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 11308
    },
    {
      "epoch": 4.518178186176588,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 11309
    },
    {
      "epoch": 4.5185777067518975,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 11310
    },
    {
      "epoch": 4.518977227327207,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 11311
    },
    {
      "epoch": 4.519376747902517,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 11312
    },
    {
      "epoch": 4.519776268477827,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 11313
    },
    {
      "epoch": 4.520175789053136,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 11314
    },
    {
      "epoch": 4.520575309628446,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 11315
    },
    {
      "epoch": 4.520974830203755,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 11316
    },
    {
      "epoch": 4.521374350779065,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 11317
    },
    {
      "epoch": 4.521773871354375,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 11318
    },
    {
      "epoch": 4.522173391929685,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 11319
    },
    {
      "epoch": 4.5225729125049945,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 11320
    },
    {
      "epoch": 4.522972433080303,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 11321
    },
    {
      "epoch": 4.523371953655613,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 11322
    },
    {
      "epoch": 4.523771474230923,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 11323
    },
    {
      "epoch": 4.524170994806233,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 11324
    },
    {
      "epoch": 4.5245705153815425,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 11325
    },
    {
      "epoch": 4.524970035956851,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 11326
    },
    {
      "epoch": 4.525369556532161,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 11327
    },
    {
      "epoch": 4.525769077107471,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 11328
    },
    {
      "epoch": 4.526168597682781,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 11329
    },
    {
      "epoch": 4.526568118258091,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 11330
    },
    {
      "epoch": 4.5269676388333995,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 11331
    },
    {
      "epoch": 4.527367159408709,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 11332
    },
    {
      "epoch": 4.527766679984019,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 11333
    },
    {
      "epoch": 4.528166200559329,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 11334
    },
    {
      "epoch": 4.528565721134639,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 11335
    },
    {
      "epoch": 4.528965241709948,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 11336
    },
    {
      "epoch": 4.529364762285257,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 11337
    },
    {
      "epoch": 4.529764282860567,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 11338
    },
    {
      "epoch": 4.530163803435877,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 11339
    },
    {
      "epoch": 4.530563324011187,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 11340
    },
    {
      "epoch": 4.5309628445864965,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 11341
    },
    {
      "epoch": 4.531362365161806,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 11342
    },
    {
      "epoch": 4.531761885737115,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 11343
    },
    {
      "epoch": 4.532161406312425,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 11344
    },
    {
      "epoch": 4.532560926887735,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 11345
    },
    {
      "epoch": 4.5329604474630445,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 11346
    },
    {
      "epoch": 4.533359968038354,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 11347
    },
    {
      "epoch": 4.533759488613663,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 11348
    },
    {
      "epoch": 4.534159009188973,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 11349
    },
    {
      "epoch": 4.534558529764283,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 11350
    },
    {
      "epoch": 4.5349580503395925,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 11351
    },
    {
      "epoch": 4.535357570914902,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 11352
    },
    {
      "epoch": 4.535757091490212,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 11353
    },
    {
      "epoch": 4.536156612065521,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0606,
      "step": 11354
    },
    {
      "epoch": 4.536556132640831,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 11355
    },
    {
      "epoch": 4.536955653216141,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 11356
    },
    {
      "epoch": 4.53735517379145,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 11357
    },
    {
      "epoch": 4.53775469436676,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 11358
    },
    {
      "epoch": 4.53815421494207,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 11359
    },
    {
      "epoch": 4.538553735517379,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 11360
    },
    {
      "epoch": 4.538953256092689,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 11361
    },
    {
      "epoch": 4.539352776667998,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 11362
    },
    {
      "epoch": 4.539752297243308,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 11363
    },
    {
      "epoch": 4.540151817818618,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 11364
    },
    {
      "epoch": 4.540551338393927,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 11365
    },
    {
      "epoch": 4.540950858969237,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 11366
    },
    {
      "epoch": 4.5413503795445465,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 11367
    },
    {
      "epoch": 4.541749900119856,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 11368
    },
    {
      "epoch": 4.542149420695166,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 11369
    },
    {
      "epoch": 4.542548941270476,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 11370
    },
    {
      "epoch": 4.542948461845785,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 11371
    },
    {
      "epoch": 4.5433479824210945,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 11372
    },
    {
      "epoch": 4.543747502996404,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 11373
    },
    {
      "epoch": 4.544147023571714,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 11374
    },
    {
      "epoch": 4.544546544147024,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 11375
    },
    {
      "epoch": 4.544946064722334,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 11376
    },
    {
      "epoch": 4.5453455852976425,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 11377
    },
    {
      "epoch": 4.545745105872952,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 11378
    },
    {
      "epoch": 4.546144626448262,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 11379
    },
    {
      "epoch": 4.546544147023572,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 11380
    },
    {
      "epoch": 4.546943667598882,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 11381
    },
    {
      "epoch": 4.547343188174191,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 11382
    },
    {
      "epoch": 4.5477427087495,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 11383
    },
    {
      "epoch": 4.54814222932481,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 11384
    },
    {
      "epoch": 4.54854174990012,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0622,
      "step": 11385
    },
    {
      "epoch": 4.54894127047543,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 11386
    },
    {
      "epoch": 4.549340791050739,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 11387
    },
    {
      "epoch": 4.549740311626048,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 11388
    },
    {
      "epoch": 4.550139832201358,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 11389
    },
    {
      "epoch": 4.550539352776668,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 11390
    },
    {
      "epoch": 4.550938873351978,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 11391
    },
    {
      "epoch": 4.551338393927288,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 11392
    },
    {
      "epoch": 4.551737914502597,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 11393
    },
    {
      "epoch": 4.552137435077906,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 11394
    },
    {
      "epoch": 4.552536955653216,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 11395
    },
    {
      "epoch": 4.552936476228526,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 11396
    },
    {
      "epoch": 4.553335996803836,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 11397
    },
    {
      "epoch": 4.553735517379145,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 11398
    },
    {
      "epoch": 4.554135037954454,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 11399
    },
    {
      "epoch": 4.554534558529764,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 11400
    },
    {
      "epoch": 4.554934079105074,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 11401
    },
    {
      "epoch": 4.555333599680384,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 11402
    },
    {
      "epoch": 4.555733120255693,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 11403
    },
    {
      "epoch": 4.556132640831002,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 11404
    },
    {
      "epoch": 4.556532161406312,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 11405
    },
    {
      "epoch": 4.556931681981622,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 11406
    },
    {
      "epoch": 4.557331202556932,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 11407
    },
    {
      "epoch": 4.5577307231322415,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 11408
    },
    {
      "epoch": 4.558130243707551,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 11409
    },
    {
      "epoch": 4.558529764282861,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 11410
    },
    {
      "epoch": 4.55892928485817,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 11411
    },
    {
      "epoch": 4.55932880543348,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 11412
    },
    {
      "epoch": 4.5597283260087895,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 11413
    },
    {
      "epoch": 4.560127846584099,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 11414
    },
    {
      "epoch": 4.560527367159409,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 11415
    },
    {
      "epoch": 4.560926887734718,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 11416
    },
    {
      "epoch": 4.561326408310028,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 11417
    },
    {
      "epoch": 4.561725928885338,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 11418
    },
    {
      "epoch": 4.562125449460647,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 11419
    },
    {
      "epoch": 4.562524970035957,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 11420
    },
    {
      "epoch": 4.562924490611266,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 11421
    },
    {
      "epoch": 4.563324011186576,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 11422
    },
    {
      "epoch": 4.563723531761886,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 11423
    },
    {
      "epoch": 4.564123052337195,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 11424
    },
    {
      "epoch": 4.564522572912505,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 11425
    },
    {
      "epoch": 4.564922093487815,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 11426
    },
    {
      "epoch": 4.565321614063124,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 11427
    },
    {
      "epoch": 4.565721134638434,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 11428
    },
    {
      "epoch": 4.566120655213743,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 11429
    },
    {
      "epoch": 4.566520175789053,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 11430
    },
    {
      "epoch": 4.566919696364363,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 11431
    },
    {
      "epoch": 4.567319216939673,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 11432
    },
    {
      "epoch": 4.567718737514982,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 11433
    },
    {
      "epoch": 4.5681182580902915,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 11434
    },
    {
      "epoch": 4.568517778665601,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 11435
    },
    {
      "epoch": 4.568917299240911,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 11436
    },
    {
      "epoch": 4.569316819816221,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 11437
    },
    {
      "epoch": 4.56971634039153,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 11438
    },
    {
      "epoch": 4.5701158609668395,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 11439
    },
    {
      "epoch": 4.570515381542149,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 11440
    },
    {
      "epoch": 4.570914902117459,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 11441
    },
    {
      "epoch": 4.571314422692769,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 11442
    },
    {
      "epoch": 4.571713943268079,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 11443
    },
    {
      "epoch": 4.572113463843388,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 11444
    },
    {
      "epoch": 4.572512984418697,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 11445
    },
    {
      "epoch": 4.572912504994007,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 11446
    },
    {
      "epoch": 4.573312025569317,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 11447
    },
    {
      "epoch": 4.573711546144627,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 11448
    },
    {
      "epoch": 4.5741110667199365,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 11449
    },
    {
      "epoch": 4.574510587295245,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 11450
    },
    {
      "epoch": 4.574910107870555,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 11451
    },
    {
      "epoch": 4.575309628445865,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 11452
    },
    {
      "epoch": 4.575709149021175,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 11453
    },
    {
      "epoch": 4.5761086695964845,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 11454
    },
    {
      "epoch": 4.5765081901717934,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 11455
    },
    {
      "epoch": 4.576907710747103,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 11456
    },
    {
      "epoch": 4.577307231322413,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 11457
    },
    {
      "epoch": 4.577706751897723,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 11458
    },
    {
      "epoch": 4.578106272473033,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 11459
    },
    {
      "epoch": 4.578505793048342,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 11460
    },
    {
      "epoch": 4.578905313623651,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 11461
    },
    {
      "epoch": 4.579304834198961,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 11462
    },
    {
      "epoch": 4.579704354774271,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 11463
    },
    {
      "epoch": 4.580103875349581,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 11464
    },
    {
      "epoch": 4.58050339592489,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 11465
    },
    {
      "epoch": 4.5809029165002,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 11466
    },
    {
      "epoch": 4.581302437075509,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 11467
    },
    {
      "epoch": 4.581701957650819,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 11468
    },
    {
      "epoch": 4.582101478226129,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 11469
    },
    {
      "epoch": 4.5825009988014385,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 11470
    },
    {
      "epoch": 4.582900519376748,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 11471
    },
    {
      "epoch": 4.583300039952057,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 11472
    },
    {
      "epoch": 4.583699560527367,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 11473
    },
    {
      "epoch": 4.584099081102677,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 11474
    },
    {
      "epoch": 4.5844986016779865,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 11475
    },
    {
      "epoch": 4.584898122253296,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 11476
    },
    {
      "epoch": 4.585297642828605,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 11477
    },
    {
      "epoch": 4.585697163403915,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 11478
    },
    {
      "epoch": 4.586096683979225,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 11479
    },
    {
      "epoch": 4.5864962045545346,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 11480
    },
    {
      "epoch": 4.586895725129844,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 11481
    },
    {
      "epoch": 4.587295245705154,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 11482
    },
    {
      "epoch": 4.587694766280464,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 11483
    },
    {
      "epoch": 4.588094286855773,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 11484
    },
    {
      "epoch": 4.588493807431083,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 11485
    },
    {
      "epoch": 4.588893328006392,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 11486
    },
    {
      "epoch": 4.589292848581702,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 11487
    },
    {
      "epoch": 4.589692369157012,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 11488
    },
    {
      "epoch": 4.590091889732321,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 11489
    },
    {
      "epoch": 4.590491410307631,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 11490
    },
    {
      "epoch": 4.59089093088294,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 11491
    },
    {
      "epoch": 4.59129045145825,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 11492
    },
    {
      "epoch": 4.59168997203356,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 11493
    },
    {
      "epoch": 4.592089492608869,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 11494
    },
    {
      "epoch": 4.592489013184179,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 11495
    },
    {
      "epoch": 4.5928885337594885,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 11496
    },
    {
      "epoch": 4.593288054334798,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 11497
    },
    {
      "epoch": 4.593687574910108,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 11498
    },
    {
      "epoch": 4.594087095485418,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 11499
    },
    {
      "epoch": 4.594486616060728,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 11500
    },
    {
      "epoch": 4.5948861366360365,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 11501
    },
    {
      "epoch": 4.595285657211346,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 11502
    },
    {
      "epoch": 4.595685177786656,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 11503
    },
    {
      "epoch": 4.596084698361966,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 11504
    },
    {
      "epoch": 4.596484218937276,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 11505
    },
    {
      "epoch": 4.5968837395125846,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 11506
    },
    {
      "epoch": 4.597283260087894,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 11507
    },
    {
      "epoch": 4.597682780663204,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 11508
    },
    {
      "epoch": 4.598082301238514,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 11509
    },
    {
      "epoch": 4.598481821813824,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 11510
    },
    {
      "epoch": 4.598881342389133,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 11511
    },
    {
      "epoch": 4.599280862964442,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0629,
      "step": 11512
    },
    {
      "epoch": 4.599680383539752,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 11513
    },
    {
      "epoch": 4.600079904115062,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 11514
    },
    {
      "epoch": 4.600479424690372,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 11515
    },
    {
      "epoch": 4.6008789452656815,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 11516
    },
    {
      "epoch": 4.60127846584099,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 11517
    },
    {
      "epoch": 4.6016779864163,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 11518
    },
    {
      "epoch": 4.60207750699161,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 11519
    },
    {
      "epoch": 4.60247702756692,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 11520
    },
    {
      "epoch": 4.60287654814223,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 11521
    },
    {
      "epoch": 4.603276068717539,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 11522
    },
    {
      "epoch": 4.603675589292848,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 11523
    },
    {
      "epoch": 4.604075109868158,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 11524
    },
    {
      "epoch": 4.604474630443468,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 11525
    },
    {
      "epoch": 4.604874151018778,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 11526
    },
    {
      "epoch": 4.605273671594087,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 11527
    },
    {
      "epoch": 4.605673192169396,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 11528
    },
    {
      "epoch": 4.606072712744706,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 11529
    },
    {
      "epoch": 4.606472233320016,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 11530
    },
    {
      "epoch": 4.606871753895326,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 11531
    },
    {
      "epoch": 4.6072712744706354,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 11532
    },
    {
      "epoch": 4.607670795045945,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 11533
    },
    {
      "epoch": 4.608070315621254,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 11534
    },
    {
      "epoch": 4.608469836196564,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 11535
    },
    {
      "epoch": 4.608869356771874,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 11536
    },
    {
      "epoch": 4.6092688773471835,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 11537
    },
    {
      "epoch": 4.609668397922493,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 11538
    },
    {
      "epoch": 4.610067918497803,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 11539
    },
    {
      "epoch": 4.610467439073112,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 11540
    },
    {
      "epoch": 4.610866959648422,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 11541
    },
    {
      "epoch": 4.6112664802237315,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 11542
    },
    {
      "epoch": 4.611666000799041,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 11543
    },
    {
      "epoch": 4.612065521374351,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 11544
    },
    {
      "epoch": 4.61246504194966,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 11545
    },
    {
      "epoch": 4.61286456252497,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 11546
    },
    {
      "epoch": 4.61326408310028,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 11547
    },
    {
      "epoch": 4.613663603675589,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 11548
    },
    {
      "epoch": 4.614063124250899,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 11549
    },
    {
      "epoch": 4.614462644826209,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0678,
      "step": 11550
    },
    {
      "epoch": 4.614862165401518,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 11551
    },
    {
      "epoch": 4.615261685976828,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 11552
    },
    {
      "epoch": 4.615661206552137,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 11553
    },
    {
      "epoch": 4.616060727127447,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 11554
    },
    {
      "epoch": 4.616460247702757,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 11555
    },
    {
      "epoch": 4.616859768278067,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 11556
    },
    {
      "epoch": 4.617259288853376,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 11557
    },
    {
      "epoch": 4.6176588094286855,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 11558
    },
    {
      "epoch": 4.618058330003995,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 11559
    },
    {
      "epoch": 4.618457850579305,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 11560
    },
    {
      "epoch": 4.618857371154615,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 11561
    },
    {
      "epoch": 4.619256891729924,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 11562
    },
    {
      "epoch": 4.6196564123052335,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 11563
    },
    {
      "epoch": 4.620055932880543,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 11564
    },
    {
      "epoch": 4.620455453455853,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 11565
    },
    {
      "epoch": 4.620854974031163,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 11566
    },
    {
      "epoch": 4.621254494606472,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 11567
    },
    {
      "epoch": 4.6216540151817815,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 11568
    },
    {
      "epoch": 4.622053535757091,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 11569
    },
    {
      "epoch": 4.622453056332401,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 11570
    },
    {
      "epoch": 4.622852576907711,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 11571
    },
    {
      "epoch": 4.623252097483021,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 11572
    },
    {
      "epoch": 4.6236516180583305,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 11573
    },
    {
      "epoch": 4.624051138633639,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 11574
    },
    {
      "epoch": 4.624450659208949,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 11575
    },
    {
      "epoch": 4.624850179784259,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 11576
    },
    {
      "epoch": 4.625249700359569,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 11577
    },
    {
      "epoch": 4.6256492209348785,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 11578
    },
    {
      "epoch": 4.626048741510187,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 11579
    },
    {
      "epoch": 4.626448262085497,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 11580
    },
    {
      "epoch": 4.626847782660807,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 11581
    },
    {
      "epoch": 4.627247303236117,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 11582
    },
    {
      "epoch": 4.627646823811427,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0639,
      "step": 11583
    },
    {
      "epoch": 4.6280463443867355,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.062,
      "step": 11584
    },
    {
      "epoch": 4.628445864962045,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 11585
    },
    {
      "epoch": 4.628845385537355,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 11586
    },
    {
      "epoch": 4.629244906112665,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 11587
    },
    {
      "epoch": 4.629644426687975,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 11588
    },
    {
      "epoch": 4.630043947263284,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 11589
    },
    {
      "epoch": 4.630443467838594,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 11590
    },
    {
      "epoch": 4.630842988413903,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 11591
    },
    {
      "epoch": 4.631242508989213,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 11592
    },
    {
      "epoch": 4.631642029564523,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 11593
    },
    {
      "epoch": 4.632041550139832,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 11594
    },
    {
      "epoch": 4.632441070715142,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 11595
    },
    {
      "epoch": 4.632840591290451,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 11596
    },
    {
      "epoch": 4.633240111865761,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 11597
    },
    {
      "epoch": 4.633639632441071,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 11598
    },
    {
      "epoch": 4.6340391530163805,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 11599
    },
    {
      "epoch": 4.63443867359169,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 11600
    },
    {
      "epoch": 4.634838194166999,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 11601
    },
    {
      "epoch": 4.635237714742309,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 11602
    },
    {
      "epoch": 4.635637235317619,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 11603
    },
    {
      "epoch": 4.6360367558929285,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 11604
    },
    {
      "epoch": 4.636436276468238,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 11605
    },
    {
      "epoch": 4.636835797043548,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 11606
    },
    {
      "epoch": 4.637235317618857,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 11607
    },
    {
      "epoch": 4.637634838194167,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 11608
    },
    {
      "epoch": 4.638034358769477,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 11609
    },
    {
      "epoch": 4.638433879344786,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 11610
    },
    {
      "epoch": 4.638833399920096,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 11611
    },
    {
      "epoch": 4.639232920495406,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 11612
    },
    {
      "epoch": 4.639632441070715,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 11613
    },
    {
      "epoch": 4.640031961646025,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 11614
    },
    {
      "epoch": 4.640431482221334,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 11615
    },
    {
      "epoch": 4.640831002796644,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 11616
    },
    {
      "epoch": 4.641230523371954,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 11617
    },
    {
      "epoch": 4.641630043947263,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 11618
    },
    {
      "epoch": 4.642029564522573,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 11619
    },
    {
      "epoch": 4.642429085097882,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 11620
    },
    {
      "epoch": 4.642828605673192,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 11621
    },
    {
      "epoch": 4.643228126248502,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 11622
    },
    {
      "epoch": 4.643627646823812,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 11623
    },
    {
      "epoch": 4.644027167399121,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 11624
    },
    {
      "epoch": 4.6444266879744305,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 11625
    },
    {
      "epoch": 4.64482620854974,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 11626
    },
    {
      "epoch": 4.64522572912505,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 11627
    },
    {
      "epoch": 4.64562524970036,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 11628
    },
    {
      "epoch": 4.64602477027567,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 11629
    },
    {
      "epoch": 4.6464242908509785,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 11630
    },
    {
      "epoch": 4.646823811426288,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 11631
    },
    {
      "epoch": 4.647223332001598,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 11632
    },
    {
      "epoch": 4.647622852576908,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 11633
    },
    {
      "epoch": 4.648022373152218,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 11634
    },
    {
      "epoch": 4.648421893727527,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 11635
    },
    {
      "epoch": 4.648821414302836,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 11636
    },
    {
      "epoch": 4.649220934878146,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 11637
    },
    {
      "epoch": 4.649620455453456,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 11638
    },
    {
      "epoch": 4.650019976028766,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 11639
    },
    {
      "epoch": 4.6504194966040755,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 11640
    },
    {
      "epoch": 4.650819017179384,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 11641
    },
    {
      "epoch": 4.651218537754694,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 11642
    },
    {
      "epoch": 4.651618058330004,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 11643
    },
    {
      "epoch": 4.652017578905314,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 11644
    },
    {
      "epoch": 4.6524170994806235,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 11645
    },
    {
      "epoch": 4.652816620055933,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 11646
    },
    {
      "epoch": 4.653216140631242,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 11647
    },
    {
      "epoch": 4.653615661206552,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 11648
    },
    {
      "epoch": 4.654015181781862,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 11649
    },
    {
      "epoch": 4.654414702357172,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 11650
    },
    {
      "epoch": 4.654814222932481,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 11651
    },
    {
      "epoch": 4.65521374350779,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 11652
    },
    {
      "epoch": 4.6556132640831,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 11653
    },
    {
      "epoch": 4.65601278465841,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 11654
    },
    {
      "epoch": 4.65641230523372,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 11655
    },
    {
      "epoch": 4.656811825809029,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 11656
    },
    {
      "epoch": 4.657211346384338,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 11657
    },
    {
      "epoch": 4.657610866959648,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 11658
    },
    {
      "epoch": 4.658010387534958,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 11659
    },
    {
      "epoch": 4.658409908110268,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 11660
    },
    {
      "epoch": 4.6588094286855775,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 11661
    },
    {
      "epoch": 4.659208949260887,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 11662
    },
    {
      "epoch": 4.659608469836197,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 11663
    },
    {
      "epoch": 4.660007990411506,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 11664
    },
    {
      "epoch": 4.660407510986816,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 11665
    },
    {
      "epoch": 4.6608070315621255,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 11666
    },
    {
      "epoch": 4.661206552137435,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 11667
    },
    {
      "epoch": 4.661606072712745,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 11668
    },
    {
      "epoch": 4.662005593288054,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 11669
    },
    {
      "epoch": 4.662405113863364,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 11670
    },
    {
      "epoch": 4.6628046344386735,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 11671
    },
    {
      "epoch": 4.663204155013983,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 11672
    },
    {
      "epoch": 4.663603675589293,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 11673
    },
    {
      "epoch": 4.664003196164602,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 11674
    },
    {
      "epoch": 4.664402716739912,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 11675
    },
    {
      "epoch": 4.664802237315222,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 11676
    },
    {
      "epoch": 4.665201757890531,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 11677
    },
    {
      "epoch": 4.665601278465841,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 11678
    },
    {
      "epoch": 4.666000799041151,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 11679
    },
    {
      "epoch": 4.666400319616461,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 11680
    },
    {
      "epoch": 4.66679984019177,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 11681
    },
    {
      "epoch": 4.667199360767079,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 11682
    },
    {
      "epoch": 4.667598881342389,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 11683
    },
    {
      "epoch": 4.667998401917699,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 11684
    },
    {
      "epoch": 4.668397922493009,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 11685
    },
    {
      "epoch": 4.668797443068318,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 11686
    },
    {
      "epoch": 4.6691969636436275,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 11687
    },
    {
      "epoch": 4.669596484218937,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 11688
    },
    {
      "epoch": 4.669996004794247,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 11689
    },
    {
      "epoch": 4.670395525369557,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 11690
    },
    {
      "epoch": 4.670795045944866,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 11691
    },
    {
      "epoch": 4.6711945665201755,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 11692
    },
    {
      "epoch": 4.671594087095485,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 11693
    },
    {
      "epoch": 4.671993607670795,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 11694
    },
    {
      "epoch": 4.672393128246105,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 11695
    },
    {
      "epoch": 4.672792648821415,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 11696
    },
    {
      "epoch": 4.6731921693967235,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 11697
    },
    {
      "epoch": 4.673591689972033,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 11698
    },
    {
      "epoch": 4.673991210547343,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 11699
    },
    {
      "epoch": 4.674390731122653,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 11700
    },
    {
      "epoch": 4.674790251697963,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 11701
    },
    {
      "epoch": 4.6751897722732725,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 11702
    },
    {
      "epoch": 4.675589292848581,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 11703
    },
    {
      "epoch": 4.675988813423891,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 11704
    },
    {
      "epoch": 4.676388333999201,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 11705
    },
    {
      "epoch": 4.676787854574511,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 11706
    },
    {
      "epoch": 4.6771873751498205,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 11707
    },
    {
      "epoch": 4.677586895725129,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 11708
    },
    {
      "epoch": 4.677986416300439,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 11709
    },
    {
      "epoch": 4.678385936875749,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 11710
    },
    {
      "epoch": 4.678785457451059,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 11711
    },
    {
      "epoch": 4.679184978026369,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 11712
    },
    {
      "epoch": 4.679584498601678,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 11713
    },
    {
      "epoch": 4.679984019176987,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 11714
    },
    {
      "epoch": 4.680383539752297,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 11715
    },
    {
      "epoch": 4.680783060327607,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 11716
    },
    {
      "epoch": 4.681182580902917,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 11717
    },
    {
      "epoch": 4.681582101478226,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 11718
    },
    {
      "epoch": 4.681981622053536,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 11719
    },
    {
      "epoch": 4.682381142628845,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 11720
    },
    {
      "epoch": 4.682780663204155,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 11721
    },
    {
      "epoch": 4.683180183779465,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 11722
    },
    {
      "epoch": 4.683579704354774,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 11723
    },
    {
      "epoch": 4.683979224930084,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 11724
    },
    {
      "epoch": 4.684378745505393,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 11725
    },
    {
      "epoch": 4.684778266080703,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 11726
    },
    {
      "epoch": 4.685177786656013,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 11727
    },
    {
      "epoch": 4.6855773072313225,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 11728
    },
    {
      "epoch": 4.685976827806632,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 11729
    },
    {
      "epoch": 4.686376348381942,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 11730
    },
    {
      "epoch": 4.686775868957251,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 11731
    },
    {
      "epoch": 4.687175389532561,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 11732
    },
    {
      "epoch": 4.6875749101078705,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 11733
    },
    {
      "epoch": 4.68797443068318,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 11734
    },
    {
      "epoch": 4.68837395125849,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 11735
    },
    {
      "epoch": 4.6887734718338,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 11736
    },
    {
      "epoch": 4.689172992409109,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 11737
    },
    {
      "epoch": 4.689572512984419,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 11738
    },
    {
      "epoch": 4.689972033559728,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 11739
    },
    {
      "epoch": 4.690371554135038,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 11740
    },
    {
      "epoch": 4.690771074710348,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 11741
    },
    {
      "epoch": 4.691170595285657,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 11742
    },
    {
      "epoch": 4.691570115860967,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 11743
    },
    {
      "epoch": 4.691969636436276,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 11744
    },
    {
      "epoch": 4.692369157011586,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 11745
    },
    {
      "epoch": 4.692768677586896,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 11746
    },
    {
      "epoch": 4.693168198162205,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 11747
    },
    {
      "epoch": 4.693567718737515,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 11748
    },
    {
      "epoch": 4.6939672393128244,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 11749
    },
    {
      "epoch": 4.694366759888134,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 11750
    },
    {
      "epoch": 4.694766280463444,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 11751
    },
    {
      "epoch": 4.695165801038754,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 11752
    },
    {
      "epoch": 4.695565321614064,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 11753
    },
    {
      "epoch": 4.6959648421893725,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 11754
    },
    {
      "epoch": 4.696364362764682,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0598,
      "step": 11755
    },
    {
      "epoch": 4.696763883339992,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 11756
    },
    {
      "epoch": 4.697163403915302,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 11757
    },
    {
      "epoch": 4.697562924490612,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 11758
    },
    {
      "epoch": 4.6979624450659205,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 11759
    },
    {
      "epoch": 4.69836196564123,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 11760
    },
    {
      "epoch": 4.69876148621654,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 11761
    },
    {
      "epoch": 4.69916100679185,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 11762
    },
    {
      "epoch": 4.69956052736716,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 11763
    },
    {
      "epoch": 4.699960047942469,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 11764
    },
    {
      "epoch": 4.700359568517778,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 11765
    },
    {
      "epoch": 4.700759089093088,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 11766
    },
    {
      "epoch": 4.701158609668398,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 11767
    },
    {
      "epoch": 4.701558130243708,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 11768
    },
    {
      "epoch": 4.7019576508190175,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 11769
    },
    {
      "epoch": 4.702357171394327,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 11770
    },
    {
      "epoch": 4.702756691969636,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 11771
    },
    {
      "epoch": 4.703156212544946,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 11772
    },
    {
      "epoch": 4.703555733120256,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 11773
    },
    {
      "epoch": 4.7039552536955656,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 11774
    },
    {
      "epoch": 4.704354774270875,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 11775
    },
    {
      "epoch": 4.704754294846184,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 11776
    },
    {
      "epoch": 4.705153815421494,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 11777
    },
    {
      "epoch": 4.705553335996804,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 11778
    },
    {
      "epoch": 4.705952856572114,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 11779
    },
    {
      "epoch": 4.706352377147423,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 11780
    },
    {
      "epoch": 4.706751897722732,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 11781
    },
    {
      "epoch": 4.707151418298042,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 11782
    },
    {
      "epoch": 4.707550938873352,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 11783
    },
    {
      "epoch": 4.707950459448662,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 11784
    },
    {
      "epoch": 4.708349980023971,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 11785
    },
    {
      "epoch": 4.708749500599281,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 11786
    },
    {
      "epoch": 4.70914902117459,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 11787
    },
    {
      "epoch": 4.7095485417499,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 11788
    },
    {
      "epoch": 4.70994806232521,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 11789
    },
    {
      "epoch": 4.7103475829005195,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 11790
    },
    {
      "epoch": 4.710747103475829,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 11791
    },
    {
      "epoch": 4.711146624051139,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 11792
    },
    {
      "epoch": 4.711546144626448,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 11793
    },
    {
      "epoch": 4.711945665201758,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 11794
    },
    {
      "epoch": 4.7123451857770675,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 11795
    },
    {
      "epoch": 4.712744706352377,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 11796
    },
    {
      "epoch": 4.713144226927687,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 11797
    },
    {
      "epoch": 4.713543747502996,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 11798
    },
    {
      "epoch": 4.713943268078306,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 11799
    },
    {
      "epoch": 4.714342788653616,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 11800
    },
    {
      "epoch": 4.714742309228925,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 11801
    },
    {
      "epoch": 4.715141829804235,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 11802
    },
    {
      "epoch": 4.715541350379545,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 11803
    },
    {
      "epoch": 4.715940870954854,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 11804
    },
    {
      "epoch": 4.716340391530164,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 11805
    },
    {
      "epoch": 4.716739912105473,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 11806
    },
    {
      "epoch": 4.717139432680783,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 11807
    },
    {
      "epoch": 4.717538953256093,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 11808
    },
    {
      "epoch": 4.717938473831403,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 11809
    },
    {
      "epoch": 4.718337994406712,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 11810
    },
    {
      "epoch": 4.718737514982021,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 11811
    },
    {
      "epoch": 4.719137035557331,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 11812
    },
    {
      "epoch": 4.719536556132641,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 11813
    },
    {
      "epoch": 4.719936076707951,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 11814
    },
    {
      "epoch": 4.72033559728326,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 11815
    },
    {
      "epoch": 4.7207351178585695,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 11816
    },
    {
      "epoch": 4.721134638433879,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 11817
    },
    {
      "epoch": 4.721534159009189,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 11818
    },
    {
      "epoch": 4.721933679584499,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 11819
    },
    {
      "epoch": 4.722333200159808,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 11820
    },
    {
      "epoch": 4.7227327207351175,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 11821
    },
    {
      "epoch": 4.723132241310427,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 11822
    },
    {
      "epoch": 4.723531761885737,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 11823
    },
    {
      "epoch": 4.723931282461047,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 11824
    },
    {
      "epoch": 4.724330803036357,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 11825
    },
    {
      "epoch": 4.7247303236116664,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 11826
    },
    {
      "epoch": 4.725129844186975,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 11827
    },
    {
      "epoch": 4.725529364762285,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 11828
    },
    {
      "epoch": 4.725928885337595,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 11829
    },
    {
      "epoch": 4.726328405912905,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 11830
    },
    {
      "epoch": 4.7267279264882145,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 11831
    },
    {
      "epoch": 4.727127447063523,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 11832
    },
    {
      "epoch": 4.727526967638833,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 11833
    },
    {
      "epoch": 4.727926488214143,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 11834
    },
    {
      "epoch": 4.728326008789453,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 11835
    },
    {
      "epoch": 4.7287255293647625,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 11836
    },
    {
      "epoch": 4.729125049940071,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 11837
    },
    {
      "epoch": 4.729524570515381,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 11838
    },
    {
      "epoch": 4.729924091090691,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 11839
    },
    {
      "epoch": 4.730323611666001,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 11840
    },
    {
      "epoch": 4.730723132241311,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 11841
    },
    {
      "epoch": 4.73112265281662,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 11842
    },
    {
      "epoch": 4.73152217339193,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 11843
    },
    {
      "epoch": 4.731921693967239,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 11844
    },
    {
      "epoch": 4.732321214542549,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 11845
    },
    {
      "epoch": 4.732720735117859,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 11846
    },
    {
      "epoch": 4.733120255693168,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 11847
    },
    {
      "epoch": 4.733519776268478,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 11848
    },
    {
      "epoch": 4.733919296843787,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 11849
    },
    {
      "epoch": 4.734318817419097,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 11850
    },
    {
      "epoch": 4.734718337994407,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 11851
    },
    {
      "epoch": 4.7351178585697165,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 11852
    },
    {
      "epoch": 4.735517379145026,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 11853
    },
    {
      "epoch": 4.735916899720335,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 11854
    },
    {
      "epoch": 4.736316420295645,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 11855
    },
    {
      "epoch": 4.736715940870955,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 11856
    },
    {
      "epoch": 4.7371154614462645,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 11857
    },
    {
      "epoch": 4.737514982021574,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 11858
    },
    {
      "epoch": 4.737914502596884,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 11859
    },
    {
      "epoch": 4.738314023172194,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 11860
    },
    {
      "epoch": 4.738713543747503,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 11861
    },
    {
      "epoch": 4.7391130643228125,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 11862
    },
    {
      "epoch": 4.739512584898122,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 11863
    },
    {
      "epoch": 4.739912105473432,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 11864
    },
    {
      "epoch": 4.740311626048742,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 11865
    },
    {
      "epoch": 4.740711146624051,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 11866
    },
    {
      "epoch": 4.741110667199361,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 11867
    },
    {
      "epoch": 4.74151018777467,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 11868
    },
    {
      "epoch": 4.74190970834998,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 11869
    },
    {
      "epoch": 4.74230922892529,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 11870
    },
    {
      "epoch": 4.742708749500599,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 11871
    },
    {
      "epoch": 4.743108270075909,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 11872
    },
    {
      "epoch": 4.743507790651218,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 11873
    },
    {
      "epoch": 4.743907311226528,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 11874
    },
    {
      "epoch": 4.744306831801838,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 11875
    },
    {
      "epoch": 4.744706352377148,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 11876
    },
    {
      "epoch": 4.745105872952457,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 11877
    },
    {
      "epoch": 4.7455053935277665,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 11878
    },
    {
      "epoch": 4.745904914103076,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 11879
    },
    {
      "epoch": 4.746304434678386,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 11880
    },
    {
      "epoch": 4.746703955253696,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 11881
    },
    {
      "epoch": 4.747103475829006,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 11882
    },
    {
      "epoch": 4.7475029964043145,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 11883
    },
    {
      "epoch": 4.747902516979624,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 11884
    },
    {
      "epoch": 4.748302037554934,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 11885
    },
    {
      "epoch": 4.748701558130244,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 11886
    },
    {
      "epoch": 4.749101078705554,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 11887
    },
    {
      "epoch": 4.7495005992808625,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 11888
    },
    {
      "epoch": 4.749900119856172,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 11889
    },
    {
      "epoch": 4.750299640431482,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 11890
    },
    {
      "epoch": 4.750699161006792,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 11891
    },
    {
      "epoch": 4.751098681582102,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 11892
    },
    {
      "epoch": 4.7514982021574115,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 11893
    },
    {
      "epoch": 4.75189772273272,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 11894
    },
    {
      "epoch": 4.75229724330803,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 11895
    },
    {
      "epoch": 4.75269676388334,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 11896
    },
    {
      "epoch": 4.75309628445865,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 11897
    },
    {
      "epoch": 4.7534958050339595,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 11898
    },
    {
      "epoch": 4.753895325609269,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 11899
    },
    {
      "epoch": 4.754294846184578,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0623,
      "step": 11900
    },
    {
      "epoch": 4.754694366759888,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 11901
    },
    {
      "epoch": 4.755093887335198,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 11902
    },
    {
      "epoch": 4.755493407910508,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 11903
    },
    {
      "epoch": 4.755892928485817,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 11904
    },
    {
      "epoch": 4.756292449061126,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 11905
    },
    {
      "epoch": 4.756691969636436,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 11906
    },
    {
      "epoch": 4.757091490211746,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 11907
    },
    {
      "epoch": 4.757491010787056,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 11908
    },
    {
      "epoch": 4.757890531362365,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 11909
    },
    {
      "epoch": 4.758290051937674,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 11910
    },
    {
      "epoch": 4.758689572512984,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 11911
    },
    {
      "epoch": 4.759089093088294,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 11912
    },
    {
      "epoch": 4.759488613663604,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 11913
    },
    {
      "epoch": 4.759888134238913,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 11914
    },
    {
      "epoch": 4.760287654814223,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 11915
    },
    {
      "epoch": 4.760687175389533,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 11916
    },
    {
      "epoch": 4.761086695964842,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 11917
    },
    {
      "epoch": 4.761486216540152,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 11918
    },
    {
      "epoch": 4.7618857371154615,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 11919
    },
    {
      "epoch": 4.762285257690771,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 11920
    },
    {
      "epoch": 4.762684778266081,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 11921
    },
    {
      "epoch": 4.76308429884139,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 11922
    },
    {
      "epoch": 4.7634838194167,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 11923
    },
    {
      "epoch": 4.7638833399920095,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 11924
    },
    {
      "epoch": 4.764282860567319,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 11925
    },
    {
      "epoch": 4.764682381142629,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 11926
    },
    {
      "epoch": 4.765081901717938,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 11927
    },
    {
      "epoch": 4.765481422293248,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 11928
    },
    {
      "epoch": 4.765880942868558,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 11929
    },
    {
      "epoch": 4.766280463443867,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 11930
    },
    {
      "epoch": 4.766679984019177,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 11931
    },
    {
      "epoch": 4.767079504594487,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 11932
    },
    {
      "epoch": 4.767479025169797,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 11933
    },
    {
      "epoch": 4.767878545745106,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 11934
    },
    {
      "epoch": 4.768278066320415,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 11935
    },
    {
      "epoch": 4.768677586895725,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 11936
    },
    {
      "epoch": 4.769077107471035,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 11937
    },
    {
      "epoch": 4.769476628046345,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 11938
    },
    {
      "epoch": 4.769876148621654,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 11939
    },
    {
      "epoch": 4.770275669196963,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 11940
    },
    {
      "epoch": 4.770675189772273,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 11941
    },
    {
      "epoch": 4.771074710347583,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 11942
    },
    {
      "epoch": 4.771474230922893,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 11943
    },
    {
      "epoch": 4.771873751498202,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 11944
    },
    {
      "epoch": 4.7722732720735115,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 11945
    },
    {
      "epoch": 4.772672792648821,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 11946
    },
    {
      "epoch": 4.773072313224131,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 11947
    },
    {
      "epoch": 4.773471833799441,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 11948
    },
    {
      "epoch": 4.773871354374751,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 11949
    },
    {
      "epoch": 4.77427087495006,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 11950
    },
    {
      "epoch": 4.774670395525369,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 11951
    },
    {
      "epoch": 4.775069916100679,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 11952
    },
    {
      "epoch": 4.775469436675989,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 11953
    },
    {
      "epoch": 4.775868957251299,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 11954
    },
    {
      "epoch": 4.7762684778266085,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 11955
    },
    {
      "epoch": 4.776667998401917,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 11956
    },
    {
      "epoch": 4.777067518977227,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 11957
    },
    {
      "epoch": 4.777467039552537,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 11958
    },
    {
      "epoch": 4.777866560127847,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 11959
    },
    {
      "epoch": 4.7782660807031565,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 11960
    },
    {
      "epoch": 4.778665601278465,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 11961
    },
    {
      "epoch": 4.779065121853775,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 11962
    },
    {
      "epoch": 4.779464642429085,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 11963
    },
    {
      "epoch": 4.779864163004395,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 11964
    },
    {
      "epoch": 4.7802636835797045,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 11965
    },
    {
      "epoch": 4.780663204155014,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 11966
    },
    {
      "epoch": 4.781062724730323,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 11967
    },
    {
      "epoch": 4.781462245305633,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 11968
    },
    {
      "epoch": 4.781861765880943,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 11969
    },
    {
      "epoch": 4.782261286456253,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 11970
    },
    {
      "epoch": 4.782660807031562,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 11971
    },
    {
      "epoch": 4.783060327606872,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 11972
    },
    {
      "epoch": 4.783459848182181,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 11973
    },
    {
      "epoch": 4.783859368757491,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 11974
    },
    {
      "epoch": 4.784258889332801,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 11975
    },
    {
      "epoch": 4.78465840990811,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 11976
    },
    {
      "epoch": 4.78505793048342,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 11977
    },
    {
      "epoch": 4.785457451058729,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 11978
    },
    {
      "epoch": 4.785856971634039,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 11979
    },
    {
      "epoch": 4.786256492209349,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 11980
    },
    {
      "epoch": 4.7866560127846585,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 11981
    },
    {
      "epoch": 4.787055533359968,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 11982
    },
    {
      "epoch": 4.787455053935278,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 11983
    },
    {
      "epoch": 4.787854574510587,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 11984
    },
    {
      "epoch": 4.788254095085897,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 11985
    },
    {
      "epoch": 4.7886536156612065,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 11986
    },
    {
      "epoch": 4.789053136236516,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 11987
    },
    {
      "epoch": 4.789452656811826,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 11988
    },
    {
      "epoch": 4.789852177387136,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 11989
    },
    {
      "epoch": 4.790251697962445,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 11990
    },
    {
      "epoch": 4.7906512185377546,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 11991
    },
    {
      "epoch": 4.791050739113064,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 11992
    },
    {
      "epoch": 4.791450259688374,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 11993
    },
    {
      "epoch": 4.791849780263684,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 11994
    },
    {
      "epoch": 4.792249300838993,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 11995
    },
    {
      "epoch": 4.792648821414303,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 11996
    },
    {
      "epoch": 4.793048341989612,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 11997
    },
    {
      "epoch": 4.793447862564922,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 11998
    },
    {
      "epoch": 4.793847383140232,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 11999
    },
    {
      "epoch": 4.794246903715541,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 12000
    },
    {
      "epoch": 4.794646424290851,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 12001
    },
    {
      "epoch": 4.79504594486616,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 12002
    },
    {
      "epoch": 4.79544546544147,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 12003
    },
    {
      "epoch": 4.79584498601678,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 12004
    },
    {
      "epoch": 4.79624450659209,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 12005
    },
    {
      "epoch": 4.7966440271674,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 12006
    },
    {
      "epoch": 4.7970435477427085,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 12007
    },
    {
      "epoch": 4.797443068318018,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 12008
    },
    {
      "epoch": 4.797842588893328,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 12009
    },
    {
      "epoch": 4.798242109468638,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 12010
    },
    {
      "epoch": 4.798641630043948,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 12011
    },
    {
      "epoch": 4.7990411506192565,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 12012
    },
    {
      "epoch": 4.799440671194566,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 12013
    },
    {
      "epoch": 4.799840191769876,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 12014
    },
    {
      "epoch": 4.800239712345186,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 12015
    },
    {
      "epoch": 4.800639232920496,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 12016
    },
    {
      "epoch": 4.801038753495805,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 12017
    },
    {
      "epoch": 4.801438274071114,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 12018
    },
    {
      "epoch": 4.801837794646424,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 12019
    },
    {
      "epoch": 4.802237315221734,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 12020
    },
    {
      "epoch": 4.802636835797044,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 12021
    },
    {
      "epoch": 4.8030363563723535,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 12022
    },
    {
      "epoch": 4.803435876947663,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 12023
    },
    {
      "epoch": 4.803835397522972,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 12024
    },
    {
      "epoch": 4.804234918098282,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 12025
    },
    {
      "epoch": 4.804634438673592,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 12026
    },
    {
      "epoch": 4.8050339592489015,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 12027
    },
    {
      "epoch": 4.805433479824211,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 12028
    },
    {
      "epoch": 4.80583300039952,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 12029
    },
    {
      "epoch": 4.80623252097483,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 12030
    },
    {
      "epoch": 4.80663204155014,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 12031
    },
    {
      "epoch": 4.80703156212545,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 12032
    },
    {
      "epoch": 4.807431082700759,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 12033
    },
    {
      "epoch": 4.807830603276068,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 12034
    },
    {
      "epoch": 4.808230123851378,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 12035
    },
    {
      "epoch": 4.808629644426688,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 12036
    },
    {
      "epoch": 4.809029165001998,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 12037
    },
    {
      "epoch": 4.809428685577307,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 12038
    },
    {
      "epoch": 4.809828206152617,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 12039
    },
    {
      "epoch": 4.810227726727927,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 12040
    },
    {
      "epoch": 4.810627247303236,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 12041
    },
    {
      "epoch": 4.811026767878546,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 12042
    },
    {
      "epoch": 4.8114262884538554,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 12043
    },
    {
      "epoch": 4.811825809029165,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 12044
    },
    {
      "epoch": 4.812225329604475,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 12045
    },
    {
      "epoch": 4.812624850179784,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 12046
    },
    {
      "epoch": 4.813024370755094,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 12047
    },
    {
      "epoch": 4.8134238913304035,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 12048
    },
    {
      "epoch": 4.813823411905713,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 12049
    },
    {
      "epoch": 4.814222932481023,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 12050
    },
    {
      "epoch": 4.814622453056332,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 12051
    },
    {
      "epoch": 4.815021973631642,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 12052
    },
    {
      "epoch": 4.8154214942069515,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 12053
    },
    {
      "epoch": 4.815821014782261,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 12054
    },
    {
      "epoch": 4.816220535357571,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 12055
    },
    {
      "epoch": 4.816620055932881,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 12056
    },
    {
      "epoch": 4.81701957650819,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 12057
    },
    {
      "epoch": 4.8174190970835,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 12058
    },
    {
      "epoch": 4.817818617658809,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 12059
    },
    {
      "epoch": 4.818218138234119,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 12060
    },
    {
      "epoch": 4.818617658809429,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0645,
      "step": 12061
    },
    {
      "epoch": 4.819017179384739,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 12062
    },
    {
      "epoch": 4.819416699960048,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 12063
    },
    {
      "epoch": 4.819816220535357,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 12064
    },
    {
      "epoch": 4.820215741110667,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 12065
    },
    {
      "epoch": 4.820615261685977,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 12066
    },
    {
      "epoch": 4.821014782261287,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 12067
    },
    {
      "epoch": 4.821414302836596,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 12068
    },
    {
      "epoch": 4.8218138234119055,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 12069
    },
    {
      "epoch": 4.822213343987215,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 12070
    },
    {
      "epoch": 4.822612864562525,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 12071
    },
    {
      "epoch": 4.823012385137835,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 12072
    },
    {
      "epoch": 4.823411905713145,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 12073
    },
    {
      "epoch": 4.8238114262884535,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 12074
    },
    {
      "epoch": 4.824210946863763,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 12075
    },
    {
      "epoch": 4.824610467439073,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 12076
    },
    {
      "epoch": 4.825009988014383,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 12077
    },
    {
      "epoch": 4.825409508589693,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 12078
    },
    {
      "epoch": 4.825809029165002,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 12079
    },
    {
      "epoch": 4.826208549740311,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 12080
    },
    {
      "epoch": 4.826608070315621,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 12081
    },
    {
      "epoch": 4.827007590890931,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 12082
    },
    {
      "epoch": 4.827407111466241,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 12083
    },
    {
      "epoch": 4.8278066320415505,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 12084
    },
    {
      "epoch": 4.828206152616859,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 12085
    },
    {
      "epoch": 4.828605673192169,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 12086
    },
    {
      "epoch": 4.829005193767479,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 12087
    },
    {
      "epoch": 4.829404714342789,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 12088
    },
    {
      "epoch": 4.8298042349180985,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 12089
    },
    {
      "epoch": 4.830203755493407,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 12090
    },
    {
      "epoch": 4.830603276068717,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 12091
    },
    {
      "epoch": 4.831002796644027,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 12092
    },
    {
      "epoch": 4.831402317219337,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 12093
    },
    {
      "epoch": 4.831801837794647,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 12094
    },
    {
      "epoch": 4.832201358369956,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 12095
    },
    {
      "epoch": 4.832600878945266,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 12096
    },
    {
      "epoch": 4.833000399520575,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 12097
    },
    {
      "epoch": 4.833399920095885,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 12098
    },
    {
      "epoch": 4.833799440671195,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 12099
    },
    {
      "epoch": 4.834198961246504,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 12100
    },
    {
      "epoch": 4.834598481821814,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 12101
    },
    {
      "epoch": 4.834998002397123,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 12102
    },
    {
      "epoch": 4.835397522972433,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 12103
    },
    {
      "epoch": 4.835797043547743,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 12104
    },
    {
      "epoch": 4.836196564123052,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 12105
    },
    {
      "epoch": 4.836596084698362,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 12106
    },
    {
      "epoch": 4.836995605273671,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 12107
    },
    {
      "epoch": 4.837395125848981,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 12108
    },
    {
      "epoch": 4.837794646424291,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 12109
    },
    {
      "epoch": 4.8381941669996005,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 12110
    },
    {
      "epoch": 4.83859368757491,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 12111
    },
    {
      "epoch": 4.83899320815022,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 12112
    },
    {
      "epoch": 4.83939272872553,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 12113
    },
    {
      "epoch": 4.839792249300839,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 12114
    },
    {
      "epoch": 4.8401917698761485,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 12115
    },
    {
      "epoch": 4.840591290451458,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 12116
    },
    {
      "epoch": 4.840990811026768,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 12117
    },
    {
      "epoch": 4.841390331602078,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 12118
    },
    {
      "epoch": 4.841789852177387,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 12119
    },
    {
      "epoch": 4.842189372752697,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 12120
    },
    {
      "epoch": 4.842588893328006,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 12121
    },
    {
      "epoch": 4.842988413903316,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 12122
    },
    {
      "epoch": 4.843387934478626,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 12123
    },
    {
      "epoch": 4.843787455053935,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 12124
    },
    {
      "epoch": 4.844186975629245,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 12125
    },
    {
      "epoch": 4.844586496204554,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 12126
    },
    {
      "epoch": 4.844986016779864,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 12127
    },
    {
      "epoch": 4.845385537355174,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 12128
    },
    {
      "epoch": 4.845785057930484,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 12129
    },
    {
      "epoch": 4.8461845785057935,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 12130
    },
    {
      "epoch": 4.846584099081102,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 12131
    },
    {
      "epoch": 4.846983619656412,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 12132
    },
    {
      "epoch": 4.847383140231722,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 12133
    },
    {
      "epoch": 4.847782660807032,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 12134
    },
    {
      "epoch": 4.848182181382342,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 12135
    },
    {
      "epoch": 4.8485817019576505,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 12136
    },
    {
      "epoch": 4.84898122253296,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 12137
    },
    {
      "epoch": 4.84938074310827,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 12138
    },
    {
      "epoch": 4.84978026368358,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 12139
    },
    {
      "epoch": 4.85017978425889,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 12140
    },
    {
      "epoch": 4.8505793048341985,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 12141
    },
    {
      "epoch": 4.850978825409508,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 12142
    },
    {
      "epoch": 4.851378345984818,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 12143
    },
    {
      "epoch": 4.851777866560128,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 12144
    },
    {
      "epoch": 4.852177387135438,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 12145
    },
    {
      "epoch": 4.8525769077107475,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 12146
    },
    {
      "epoch": 4.852976428286056,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 12147
    },
    {
      "epoch": 4.853375948861366,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 12148
    },
    {
      "epoch": 4.853775469436676,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 12149
    },
    {
      "epoch": 4.854174990011986,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 12150
    },
    {
      "epoch": 4.8545745105872955,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 12151
    },
    {
      "epoch": 4.854974031162605,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 12152
    },
    {
      "epoch": 4.855373551737914,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 12153
    },
    {
      "epoch": 4.855773072313224,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 12154
    },
    {
      "epoch": 4.856172592888534,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 12155
    },
    {
      "epoch": 4.8565721134638435,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 12156
    },
    {
      "epoch": 4.856971634039153,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 12157
    },
    {
      "epoch": 4.857371154614462,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 12158
    },
    {
      "epoch": 4.857770675189772,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 12159
    },
    {
      "epoch": 4.858170195765082,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 12160
    },
    {
      "epoch": 4.858569716340392,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 12161
    },
    {
      "epoch": 4.858969236915701,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 12162
    },
    {
      "epoch": 4.859368757491011,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 12163
    },
    {
      "epoch": 4.85976827806632,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 12164
    },
    {
      "epoch": 4.86016779864163,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 12165
    },
    {
      "epoch": 4.86056731921694,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 12166
    },
    {
      "epoch": 4.860966839792249,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 12167
    },
    {
      "epoch": 4.861366360367559,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 12168
    },
    {
      "epoch": 4.861765880942869,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 12169
    },
    {
      "epoch": 4.862165401518178,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 12170
    },
    {
      "epoch": 4.862564922093488,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 12171
    },
    {
      "epoch": 4.8629644426687975,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 12172
    },
    {
      "epoch": 4.863363963244107,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 12173
    },
    {
      "epoch": 4.863763483819417,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 12174
    },
    {
      "epoch": 4.864163004394726,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 12175
    },
    {
      "epoch": 4.864562524970036,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 12176
    },
    {
      "epoch": 4.8649620455453455,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 12177
    },
    {
      "epoch": 4.865361566120655,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 12178
    },
    {
      "epoch": 4.865761086695965,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 12179
    },
    {
      "epoch": 4.866160607271274,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 12180
    },
    {
      "epoch": 4.866560127846584,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 12181
    },
    {
      "epoch": 4.8669596484218935,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 12182
    },
    {
      "epoch": 4.867359168997203,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 12183
    },
    {
      "epoch": 4.867758689572513,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 12184
    },
    {
      "epoch": 4.868158210147823,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 12185
    },
    {
      "epoch": 4.868557730723133,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 12186
    },
    {
      "epoch": 4.868957251298442,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 12187
    },
    {
      "epoch": 4.869356771873751,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 12188
    },
    {
      "epoch": 4.869756292449061,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 12189
    },
    {
      "epoch": 4.870155813024371,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 12190
    },
    {
      "epoch": 4.870555333599681,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 12191
    },
    {
      "epoch": 4.87095485417499,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 12192
    },
    {
      "epoch": 4.871354374750299,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 12193
    },
    {
      "epoch": 4.871753895325609,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 12194
    },
    {
      "epoch": 4.872153415900919,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 12195
    },
    {
      "epoch": 4.872552936476229,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 12196
    },
    {
      "epoch": 4.872952457051538,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 12197
    },
    {
      "epoch": 4.8733519776268475,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 12198
    },
    {
      "epoch": 4.873751498202157,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0655,
      "step": 12199
    },
    {
      "epoch": 4.874151018777467,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 12200
    },
    {
      "epoch": 4.874550539352777,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 12201
    },
    {
      "epoch": 4.874950059928087,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 12202
    },
    {
      "epoch": 4.875349580503396,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 12203
    },
    {
      "epoch": 4.875749101078705,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 12204
    },
    {
      "epoch": 4.876148621654015,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 12205
    },
    {
      "epoch": 4.876548142229325,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 12206
    },
    {
      "epoch": 4.876947662804635,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 12207
    },
    {
      "epoch": 4.877347183379944,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 12208
    },
    {
      "epoch": 4.877746703955253,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 12209
    },
    {
      "epoch": 4.878146224530563,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 12210
    },
    {
      "epoch": 4.878545745105873,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 12211
    },
    {
      "epoch": 4.878945265681183,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 12212
    },
    {
      "epoch": 4.8793447862564925,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 12213
    },
    {
      "epoch": 4.879744306831801,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 12214
    },
    {
      "epoch": 4.880143827407111,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 12215
    },
    {
      "epoch": 4.880543347982421,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 12216
    },
    {
      "epoch": 4.880942868557731,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 12217
    },
    {
      "epoch": 4.8813423891330405,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.066,
      "step": 12218
    },
    {
      "epoch": 4.88174190970835,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 12219
    },
    {
      "epoch": 4.88214143028366,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 12220
    },
    {
      "epoch": 4.882540950858969,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 12221
    },
    {
      "epoch": 4.882940471434279,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 12222
    },
    {
      "epoch": 4.883339992009589,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 12223
    },
    {
      "epoch": 4.883739512584898,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 12224
    },
    {
      "epoch": 4.884139033160208,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 12225
    },
    {
      "epoch": 4.884538553735517,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 12226
    },
    {
      "epoch": 4.884938074310827,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 12227
    },
    {
      "epoch": 4.885337594886137,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 12228
    },
    {
      "epoch": 4.885737115461446,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 12229
    },
    {
      "epoch": 4.886136636036756,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 12230
    },
    {
      "epoch": 4.886536156612065,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 12231
    },
    {
      "epoch": 4.886935677187375,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 12232
    },
    {
      "epoch": 4.887335197762685,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 12233
    },
    {
      "epoch": 4.887734718337994,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 12234
    },
    {
      "epoch": 4.888134238913304,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 12235
    },
    {
      "epoch": 4.888533759488614,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 12236
    },
    {
      "epoch": 4.888933280063923,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 12237
    },
    {
      "epoch": 4.889332800639233,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 12238
    },
    {
      "epoch": 4.8897323212145425,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 12239
    },
    {
      "epoch": 4.890131841789852,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 12240
    },
    {
      "epoch": 4.890531362365162,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 12241
    },
    {
      "epoch": 4.890930882940472,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 12242
    },
    {
      "epoch": 4.891330403515781,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 12243
    },
    {
      "epoch": 4.8917299240910905,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 12244
    },
    {
      "epoch": 4.8921294446664,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 12245
    },
    {
      "epoch": 4.89252896524171,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 12246
    },
    {
      "epoch": 4.89292848581702,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 12247
    },
    {
      "epoch": 4.893328006392329,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 12248
    },
    {
      "epoch": 4.893727526967639,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 12249
    },
    {
      "epoch": 4.894127047542948,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 12250
    },
    {
      "epoch": 4.894526568118258,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 12251
    },
    {
      "epoch": 4.894926088693568,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 12252
    },
    {
      "epoch": 4.895325609268878,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 12253
    },
    {
      "epoch": 4.895725129844187,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 12254
    },
    {
      "epoch": 4.896124650419496,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 12255
    },
    {
      "epoch": 4.896524170994806,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 12256
    },
    {
      "epoch": 4.896923691570116,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 12257
    },
    {
      "epoch": 4.897323212145426,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 12258
    },
    {
      "epoch": 4.8977227327207356,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 12259
    },
    {
      "epoch": 4.8981222532960444,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 12260
    },
    {
      "epoch": 4.898521773871354,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 12261
    },
    {
      "epoch": 4.898921294446664,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 12262
    },
    {
      "epoch": 4.899320815021974,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 12263
    },
    {
      "epoch": 4.899720335597284,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 12264
    },
    {
      "epoch": 4.9001198561725925,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 12265
    },
    {
      "epoch": 4.900519376747902,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 12266
    },
    {
      "epoch": 4.900918897323212,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 12267
    },
    {
      "epoch": 4.901318417898522,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 12268
    },
    {
      "epoch": 4.901717938473832,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 12269
    },
    {
      "epoch": 4.9021174590491405,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 12270
    },
    {
      "epoch": 4.90251697962445,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 12271
    },
    {
      "epoch": 4.90291650019976,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 12272
    },
    {
      "epoch": 4.90331602077507,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 12273
    },
    {
      "epoch": 4.90371554135038,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 12274
    },
    {
      "epoch": 4.9041150619256895,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 12275
    },
    {
      "epoch": 4.904514582500999,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 12276
    },
    {
      "epoch": 4.904914103076308,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 12277
    },
    {
      "epoch": 4.905313623651618,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 12278
    },
    {
      "epoch": 4.905713144226928,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 12279
    },
    {
      "epoch": 4.9061126648022375,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 12280
    },
    {
      "epoch": 4.906512185377547,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 12281
    },
    {
      "epoch": 4.906911705952856,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 12282
    },
    {
      "epoch": 4.907311226528166,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 12283
    },
    {
      "epoch": 4.907710747103476,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 12284
    },
    {
      "epoch": 4.9081102676787856,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 12285
    },
    {
      "epoch": 4.908509788254095,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 12286
    },
    {
      "epoch": 4.908909308829404,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 12287
    },
    {
      "epoch": 4.909308829404714,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 12288
    },
    {
      "epoch": 4.909708349980024,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 12289
    },
    {
      "epoch": 4.910107870555334,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 12290
    },
    {
      "epoch": 4.910507391130643,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 12291
    },
    {
      "epoch": 4.910906911705953,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 12292
    },
    {
      "epoch": 4.911306432281263,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 12293
    },
    {
      "epoch": 4.911705952856572,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 12294
    },
    {
      "epoch": 4.912105473431882,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 12295
    },
    {
      "epoch": 4.912504994007191,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 12296
    },
    {
      "epoch": 4.912904514582501,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 12297
    },
    {
      "epoch": 4.913304035157811,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 12298
    },
    {
      "epoch": 4.91370355573312,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 12299
    },
    {
      "epoch": 4.91410307630843,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 12300
    },
    {
      "epoch": 4.9145025968837395,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 12301
    },
    {
      "epoch": 4.914902117459049,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 12302
    },
    {
      "epoch": 4.915301638034359,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 12303
    },
    {
      "epoch": 4.915701158609668,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 12304
    },
    {
      "epoch": 4.916100679184978,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 12305
    },
    {
      "epoch": 4.9165001997602875,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 12306
    },
    {
      "epoch": 4.916899720335597,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 12307
    },
    {
      "epoch": 4.917299240910907,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 12308
    },
    {
      "epoch": 4.917698761486217,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 12309
    },
    {
      "epoch": 4.918098282061527,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 12310
    },
    {
      "epoch": 4.918497802636836,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 12311
    },
    {
      "epoch": 4.918897323212145,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 12312
    },
    {
      "epoch": 4.919296843787455,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 12313
    },
    {
      "epoch": 4.919696364362765,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 12314
    },
    {
      "epoch": 4.920095884938075,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 12315
    },
    {
      "epoch": 4.920495405513384,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 12316
    },
    {
      "epoch": 4.920894926088693,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 12317
    },
    {
      "epoch": 4.921294446664003,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 12318
    },
    {
      "epoch": 4.921693967239313,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 12319
    },
    {
      "epoch": 4.922093487814623,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 12320
    },
    {
      "epoch": 4.922493008389932,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 12321
    },
    {
      "epoch": 4.922892528965241,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 12322
    },
    {
      "epoch": 4.923292049540551,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 12323
    },
    {
      "epoch": 4.923691570115861,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 12324
    },
    {
      "epoch": 4.924091090691171,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 12325
    },
    {
      "epoch": 4.924490611266481,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 12326
    },
    {
      "epoch": 4.9248901318417895,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 12327
    },
    {
      "epoch": 4.925289652417099,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 12328
    },
    {
      "epoch": 4.925689172992409,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 12329
    },
    {
      "epoch": 4.926088693567719,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 12330
    },
    {
      "epoch": 4.926488214143029,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 12331
    },
    {
      "epoch": 4.926887734718338,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 12332
    },
    {
      "epoch": 4.927287255293647,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 12333
    },
    {
      "epoch": 4.927686775868957,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 12334
    },
    {
      "epoch": 4.928086296444267,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 12335
    },
    {
      "epoch": 4.928485817019577,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 12336
    },
    {
      "epoch": 4.9288853375948865,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 12337
    },
    {
      "epoch": 4.929284858170195,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 12338
    },
    {
      "epoch": 4.929684378745505,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 12339
    },
    {
      "epoch": 4.930083899320815,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 12340
    },
    {
      "epoch": 4.930483419896125,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 12341
    },
    {
      "epoch": 4.9308829404714345,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 12342
    },
    {
      "epoch": 4.931282461046744,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 12343
    },
    {
      "epoch": 4.931681981622053,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 12344
    },
    {
      "epoch": 4.932081502197363,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 12345
    },
    {
      "epoch": 4.932481022772673,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 12346
    },
    {
      "epoch": 4.9328805433479825,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 12347
    },
    {
      "epoch": 4.933280063923292,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 12348
    },
    {
      "epoch": 4.933679584498602,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 12349
    },
    {
      "epoch": 4.934079105073911,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 12350
    },
    {
      "epoch": 4.934478625649221,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 12351
    },
    {
      "epoch": 4.934878146224531,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 12352
    },
    {
      "epoch": 4.93527766679984,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 12353
    },
    {
      "epoch": 4.93567718737515,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 12354
    },
    {
      "epoch": 4.936076707950459,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 12355
    },
    {
      "epoch": 4.936476228525769,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 12356
    },
    {
      "epoch": 4.936875749101079,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 12357
    },
    {
      "epoch": 4.937275269676388,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 12358
    },
    {
      "epoch": 4.937674790251698,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 12359
    },
    {
      "epoch": 4.938074310827007,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 12360
    },
    {
      "epoch": 4.938473831402317,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 12361
    },
    {
      "epoch": 4.938873351977627,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 12362
    },
    {
      "epoch": 4.9392728725529365,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 12363
    },
    {
      "epoch": 4.939672393128246,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 12364
    },
    {
      "epoch": 4.940071913703556,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 12365
    },
    {
      "epoch": 4.940471434278866,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 12366
    },
    {
      "epoch": 4.940870954854175,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 12367
    },
    {
      "epoch": 4.9412704754294845,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 12368
    },
    {
      "epoch": 4.941669996004794,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 12369
    },
    {
      "epoch": 4.942069516580104,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 12370
    },
    {
      "epoch": 4.942469037155414,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 12371
    },
    {
      "epoch": 4.942868557730723,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 12372
    },
    {
      "epoch": 4.9432680783060325,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 12373
    },
    {
      "epoch": 4.943667598881342,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 12374
    },
    {
      "epoch": 4.944067119456652,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 12375
    },
    {
      "epoch": 4.944466640031962,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 12376
    },
    {
      "epoch": 4.944866160607271,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 12377
    },
    {
      "epoch": 4.945265681182581,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 12378
    },
    {
      "epoch": 4.94566520175789,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 12379
    },
    {
      "epoch": 4.9460647223332,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 12380
    },
    {
      "epoch": 4.94646424290851,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 12381
    },
    {
      "epoch": 4.94686376348382,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 12382
    },
    {
      "epoch": 4.9472632840591295,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 12383
    },
    {
      "epoch": 4.947662804634438,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 12384
    },
    {
      "epoch": 4.948062325209748,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 12385
    },
    {
      "epoch": 4.948461845785058,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 12386
    },
    {
      "epoch": 4.948861366360368,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 12387
    },
    {
      "epoch": 4.949260886935678,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 12388
    },
    {
      "epoch": 4.9496604075109865,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 12389
    },
    {
      "epoch": 4.950059928086296,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 12390
    },
    {
      "epoch": 4.950459448661606,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 12391
    },
    {
      "epoch": 4.950858969236916,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 12392
    },
    {
      "epoch": 4.951258489812226,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 12393
    },
    {
      "epoch": 4.9516580103875345,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 12394
    },
    {
      "epoch": 4.952057530962844,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 12395
    },
    {
      "epoch": 4.952457051538154,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 12396
    },
    {
      "epoch": 4.952856572113464,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 12397
    },
    {
      "epoch": 4.953256092688774,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 12398
    },
    {
      "epoch": 4.953655613264083,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 12399
    },
    {
      "epoch": 4.954055133839393,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 12400
    },
    {
      "epoch": 4.954454654414702,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 12401
    },
    {
      "epoch": 4.954854174990012,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 12402
    },
    {
      "epoch": 4.955253695565322,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 12403
    },
    {
      "epoch": 4.9556532161406315,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 12404
    },
    {
      "epoch": 4.956052736715941,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 12405
    },
    {
      "epoch": 4.95645225729125,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 12406
    },
    {
      "epoch": 4.95685177786656,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 12407
    },
    {
      "epoch": 4.95725129844187,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 12408
    },
    {
      "epoch": 4.9576508190171795,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0621,
      "step": 12409
    },
    {
      "epoch": 4.958050339592489,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 12410
    },
    {
      "epoch": 4.958449860167798,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 12411
    },
    {
      "epoch": 4.958849380743108,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 12412
    },
    {
      "epoch": 4.959248901318418,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 12413
    },
    {
      "epoch": 4.959648421893728,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 12414
    },
    {
      "epoch": 4.960047942469037,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 12415
    },
    {
      "epoch": 4.960447463044347,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 12416
    },
    {
      "epoch": 4.960846983619656,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 12417
    },
    {
      "epoch": 4.961246504194966,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 12418
    },
    {
      "epoch": 4.961646024770276,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 12419
    },
    {
      "epoch": 4.962045545345585,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 12420
    },
    {
      "epoch": 4.962445065920895,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 12421
    },
    {
      "epoch": 4.962844586496205,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 12422
    },
    {
      "epoch": 4.963244107071514,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 12423
    },
    {
      "epoch": 4.963643627646824,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 12424
    },
    {
      "epoch": 4.964043148222133,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 12425
    },
    {
      "epoch": 4.964442668797443,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 12426
    },
    {
      "epoch": 4.964842189372753,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 12427
    },
    {
      "epoch": 4.965241709948062,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 12428
    },
    {
      "epoch": 4.965641230523372,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 12429
    },
    {
      "epoch": 4.9660407510986815,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0596,
      "step": 12430
    },
    {
      "epoch": 4.966440271673991,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 12431
    },
    {
      "epoch": 4.966839792249301,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 12432
    },
    {
      "epoch": 4.967239312824611,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 12433
    },
    {
      "epoch": 4.96763883339992,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 12434
    },
    {
      "epoch": 4.9680383539752295,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 12435
    },
    {
      "epoch": 4.968437874550539,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 12436
    },
    {
      "epoch": 4.968837395125849,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 12437
    },
    {
      "epoch": 4.969236915701159,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 12438
    },
    {
      "epoch": 4.969636436276469,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 12439
    },
    {
      "epoch": 4.970035956851778,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 12440
    },
    {
      "epoch": 4.970435477427087,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 12441
    },
    {
      "epoch": 4.970834998002397,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 12442
    },
    {
      "epoch": 4.971234518577707,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 12443
    },
    {
      "epoch": 4.971634039153017,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 12444
    },
    {
      "epoch": 4.972033559728326,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 12445
    },
    {
      "epoch": 4.972433080303635,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 12446
    },
    {
      "epoch": 4.972832600878945,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 12447
    },
    {
      "epoch": 4.973232121454255,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 12448
    },
    {
      "epoch": 4.973631642029565,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 12449
    },
    {
      "epoch": 4.974031162604874,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 12450
    },
    {
      "epoch": 4.974430683180183,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 12451
    },
    {
      "epoch": 4.974830203755493,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 12452
    },
    {
      "epoch": 4.975229724330803,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 12453
    },
    {
      "epoch": 4.975629244906113,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 12454
    },
    {
      "epoch": 4.976028765481423,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 12455
    },
    {
      "epoch": 4.976428286056732,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 12456
    },
    {
      "epoch": 4.976827806632041,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 12457
    },
    {
      "epoch": 4.977227327207351,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 12458
    },
    {
      "epoch": 4.977626847782661,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 12459
    },
    {
      "epoch": 4.978026368357971,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 12460
    },
    {
      "epoch": 4.97842588893328,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 12461
    },
    {
      "epoch": 4.978825409508589,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 12462
    },
    {
      "epoch": 4.979224930083899,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 12463
    },
    {
      "epoch": 4.979624450659209,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 12464
    },
    {
      "epoch": 4.980023971234519,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 12465
    },
    {
      "epoch": 4.9804234918098285,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 12466
    },
    {
      "epoch": 4.980823012385137,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 12467
    },
    {
      "epoch": 4.981222532960447,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 12468
    },
    {
      "epoch": 4.981622053535757,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 12469
    },
    {
      "epoch": 4.982021574111067,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 12470
    },
    {
      "epoch": 4.9824210946863765,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 12471
    },
    {
      "epoch": 4.982820615261686,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 12472
    },
    {
      "epoch": 4.983220135836996,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 12473
    },
    {
      "epoch": 4.983619656412305,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 12474
    },
    {
      "epoch": 4.984019176987615,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 12475
    },
    {
      "epoch": 4.9844186975629245,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 12476
    },
    {
      "epoch": 4.984818218138234,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 12477
    },
    {
      "epoch": 4.985217738713544,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 12478
    },
    {
      "epoch": 4.985617259288853,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 12479
    },
    {
      "epoch": 4.986016779864163,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 12480
    },
    {
      "epoch": 4.986416300439473,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 12481
    },
    {
      "epoch": 4.986815821014782,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 12482
    },
    {
      "epoch": 4.987215341590092,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 12483
    },
    {
      "epoch": 4.987614862165401,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 12484
    },
    {
      "epoch": 4.988014382740711,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 12485
    },
    {
      "epoch": 4.988413903316021,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 12486
    },
    {
      "epoch": 4.98881342389133,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 12487
    },
    {
      "epoch": 4.98921294446664,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 12488
    },
    {
      "epoch": 4.98961246504195,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 12489
    },
    {
      "epoch": 4.99001198561726,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 12490
    },
    {
      "epoch": 4.990411506192569,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 12491
    },
    {
      "epoch": 4.9908110267678785,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 12492
    },
    {
      "epoch": 4.991210547343188,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 12493
    },
    {
      "epoch": 4.991610067918498,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 12494
    },
    {
      "epoch": 4.992009588493808,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0652,
      "step": 12495
    },
    {
      "epoch": 4.992409109069117,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 12496
    },
    {
      "epoch": 4.9928086296444265,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 12497
    },
    {
      "epoch": 4.993208150219736,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 12498
    },
    {
      "epoch": 4.993607670795046,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 12499
    },
    {
      "epoch": 4.994007191370356,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 12500
    },
    {
      "epoch": 4.994406711945665,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 12501
    },
    {
      "epoch": 4.9948062325209746,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 12502
    },
    {
      "epoch": 4.995205753096284,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 12503
    },
    {
      "epoch": 4.995605273671594,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 12504
    },
    {
      "epoch": 4.996004794246904,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 12505
    },
    {
      "epoch": 4.996404314822214,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 12506
    },
    {
      "epoch": 4.996803835397523,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 12507
    },
    {
      "epoch": 4.997203355972832,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 12508
    },
    {
      "epoch": 4.997602876548142,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 12509
    },
    {
      "epoch": 4.998002397123452,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 12510
    },
    {
      "epoch": 4.998401917698762,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 12511
    },
    {
      "epoch": 4.9988014382740715,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 12512
    },
    {
      "epoch": 4.99920095884938,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 12513
    },
    {
      "epoch": 4.99960047942469,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 12514
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 12515
    },
    {
      "epoch": 5.00039952057531,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 12516
    },
    {
      "epoch": 5.00079904115062,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 12517
    },
    {
      "epoch": 5.0011985617259285,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 12518
    },
    {
      "epoch": 5.001598082301238,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 12519
    },
    {
      "epoch": 5.001997602876548,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 12520
    },
    {
      "epoch": 5.002397123451858,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 12521
    },
    {
      "epoch": 5.002796644027168,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 12522
    },
    {
      "epoch": 5.003196164602477,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 12523
    },
    {
      "epoch": 5.003595685177786,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 12524
    },
    {
      "epoch": 5.003995205753096,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 12525
    },
    {
      "epoch": 5.004394726328406,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 12526
    },
    {
      "epoch": 5.004794246903716,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 12527
    },
    {
      "epoch": 5.0051937674790254,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 12528
    },
    {
      "epoch": 5.005593288054335,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 12529
    },
    {
      "epoch": 5.005992808629644,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 12530
    },
    {
      "epoch": 5.006392329204954,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 12531
    },
    {
      "epoch": 5.006791849780264,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 12532
    },
    {
      "epoch": 5.0071913703555735,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 12533
    },
    {
      "epoch": 5.007590890930883,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 12534
    },
    {
      "epoch": 5.007990411506192,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 12535
    },
    {
      "epoch": 5.008389932081502,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 12536
    },
    {
      "epoch": 5.008789452656812,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 12537
    },
    {
      "epoch": 5.0091889732321215,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 12538
    },
    {
      "epoch": 5.009588493807431,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 12539
    },
    {
      "epoch": 5.009988014382741,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 12540
    },
    {
      "epoch": 5.01038753495805,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 12541
    },
    {
      "epoch": 5.01078705553336,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 12542
    },
    {
      "epoch": 5.01118657610867,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 12543
    },
    {
      "epoch": 5.011586096683979,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 12544
    },
    {
      "epoch": 5.011985617259289,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0607,
      "step": 12545
    },
    {
      "epoch": 5.012385137834598,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 12546
    },
    {
      "epoch": 5.012784658409908,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 12547
    },
    {
      "epoch": 5.013184178985218,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 12548
    },
    {
      "epoch": 5.013583699560527,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 12549
    },
    {
      "epoch": 5.013983220135837,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 12550
    },
    {
      "epoch": 5.014382740711147,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 12551
    },
    {
      "epoch": 5.014782261286456,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 12552
    },
    {
      "epoch": 5.015181781861766,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 12553
    },
    {
      "epoch": 5.0155813024370755,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 12554
    },
    {
      "epoch": 5.015980823012385,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 12555
    },
    {
      "epoch": 5.016380343587695,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 12556
    },
    {
      "epoch": 5.016779864163005,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 12557
    },
    {
      "epoch": 5.017179384738314,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 12558
    },
    {
      "epoch": 5.0175789053136235,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 12559
    },
    {
      "epoch": 5.017978425888933,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 12560
    },
    {
      "epoch": 5.018377946464243,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 12561
    },
    {
      "epoch": 5.018777467039553,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 12562
    },
    {
      "epoch": 5.019176987614862,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 12563
    },
    {
      "epoch": 5.0195765081901715,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 12564
    },
    {
      "epoch": 5.019976028765481,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 12565
    },
    {
      "epoch": 5.020375549340791,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 12566
    },
    {
      "epoch": 5.020775069916101,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 12567
    },
    {
      "epoch": 5.021174590491411,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 12568
    },
    {
      "epoch": 5.02157411106672,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 12569
    },
    {
      "epoch": 5.021973631642029,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 12570
    },
    {
      "epoch": 5.022373152217339,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 12571
    },
    {
      "epoch": 5.022772672792649,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 12572
    },
    {
      "epoch": 5.023172193367959,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 12573
    },
    {
      "epoch": 5.0235717139432685,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 12574
    },
    {
      "epoch": 5.023971234518577,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 12575
    },
    {
      "epoch": 5.024370755093887,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 12576
    },
    {
      "epoch": 5.024770275669197,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 12577
    },
    {
      "epoch": 5.025169796244507,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 12578
    },
    {
      "epoch": 5.025569316819817,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 12579
    },
    {
      "epoch": 5.0259688373951255,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 12580
    },
    {
      "epoch": 5.026368357970435,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 12581
    },
    {
      "epoch": 5.026767878545745,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 12582
    },
    {
      "epoch": 5.027167399121055,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 12583
    },
    {
      "epoch": 5.027566919696365,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 12584
    },
    {
      "epoch": 5.027966440271674,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 12585
    },
    {
      "epoch": 5.028365960846983,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 12586
    },
    {
      "epoch": 5.028765481422293,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 12587
    },
    {
      "epoch": 5.029165001997603,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0611,
      "step": 12588
    },
    {
      "epoch": 5.029564522572913,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 12589
    },
    {
      "epoch": 5.029964043148222,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 12590
    },
    {
      "epoch": 5.030363563723531,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 12591
    },
    {
      "epoch": 5.030763084298841,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 12592
    },
    {
      "epoch": 5.031162604874151,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 12593
    },
    {
      "epoch": 5.031562125449461,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 12594
    },
    {
      "epoch": 5.0319616460247705,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 12595
    },
    {
      "epoch": 5.03236116660008,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 12596
    },
    {
      "epoch": 5.032760687175389,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 12597
    },
    {
      "epoch": 5.033160207750699,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 12598
    },
    {
      "epoch": 5.033559728326009,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 12599
    },
    {
      "epoch": 5.0339592489013185,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 12600
    },
    {
      "epoch": 5.034358769476628,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 12601
    },
    {
      "epoch": 5.034758290051938,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 12602
    },
    {
      "epoch": 5.035157810627247,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 12603
    },
    {
      "epoch": 5.035557331202557,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 12604
    },
    {
      "epoch": 5.035956851777867,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 12605
    },
    {
      "epoch": 5.036356372353176,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 12606
    },
    {
      "epoch": 5.036755892928486,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 12607
    },
    {
      "epoch": 5.037155413503795,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 12608
    },
    {
      "epoch": 5.037554934079105,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 12609
    },
    {
      "epoch": 5.037954454654415,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 12610
    },
    {
      "epoch": 5.038353975229724,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 12611
    },
    {
      "epoch": 5.038753495805034,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 12612
    },
    {
      "epoch": 5.039153016380344,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 12613
    },
    {
      "epoch": 5.039552536955653,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 12614
    },
    {
      "epoch": 5.039952057530963,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 12615
    },
    {
      "epoch": 5.040351578106272,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 12616
    },
    {
      "epoch": 5.040751098681582,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 12617
    },
    {
      "epoch": 5.041150619256892,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 12618
    },
    {
      "epoch": 5.041550139832202,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 12619
    },
    {
      "epoch": 5.041949660407511,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 12620
    },
    {
      "epoch": 5.0423491809828205,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 12621
    },
    {
      "epoch": 5.04274870155813,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 12622
    },
    {
      "epoch": 5.04314822213344,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 12623
    },
    {
      "epoch": 5.04354774270875,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 12624
    },
    {
      "epoch": 5.043947263284059,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 12625
    },
    {
      "epoch": 5.0443467838593685,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 12626
    },
    {
      "epoch": 5.044746304434678,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 12627
    },
    {
      "epoch": 5.045145825009988,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 12628
    },
    {
      "epoch": 5.045545345585298,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 12629
    },
    {
      "epoch": 5.045944866160608,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 12630
    },
    {
      "epoch": 5.046344386735917,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 12631
    },
    {
      "epoch": 5.046743907311226,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 12632
    },
    {
      "epoch": 5.047143427886536,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 12633
    },
    {
      "epoch": 5.047542948461846,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 12634
    },
    {
      "epoch": 5.047942469037156,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 12635
    },
    {
      "epoch": 5.048341989612465,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 12636
    },
    {
      "epoch": 5.048741510187774,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 12637
    },
    {
      "epoch": 5.049141030763084,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 12638
    },
    {
      "epoch": 5.049540551338394,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 12639
    },
    {
      "epoch": 5.049940071913704,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 12640
    },
    {
      "epoch": 5.0503395924890135,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 12641
    },
    {
      "epoch": 5.050739113064322,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 12642
    },
    {
      "epoch": 5.051138633639632,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 12643
    },
    {
      "epoch": 5.051538154214942,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 12644
    },
    {
      "epoch": 5.051937674790252,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 12645
    },
    {
      "epoch": 5.052337195365562,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 12646
    },
    {
      "epoch": 5.052736715940871,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 12647
    },
    {
      "epoch": 5.05313623651618,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 12648
    },
    {
      "epoch": 5.05353575709149,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 12649
    },
    {
      "epoch": 5.0539352776668,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 12650
    },
    {
      "epoch": 5.05433479824211,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 12651
    },
    {
      "epoch": 5.054734318817419,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 12652
    },
    {
      "epoch": 5.055133839392728,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 12653
    },
    {
      "epoch": 5.055533359968038,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 12654
    },
    {
      "epoch": 5.055932880543348,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 12655
    },
    {
      "epoch": 5.056332401118658,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 12656
    },
    {
      "epoch": 5.0567319216939675,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 12657
    },
    {
      "epoch": 5.057131442269277,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 12658
    },
    {
      "epoch": 5.057530962844586,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 12659
    },
    {
      "epoch": 5.057930483419896,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 12660
    },
    {
      "epoch": 5.058330003995206,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 12661
    },
    {
      "epoch": 5.0587295245705155,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 12662
    },
    {
      "epoch": 5.059129045145825,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 12663
    },
    {
      "epoch": 5.059528565721135,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 12664
    },
    {
      "epoch": 5.059928086296444,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 12665
    },
    {
      "epoch": 5.060327606871754,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 12666
    },
    {
      "epoch": 5.0607271274470635,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 12667
    },
    {
      "epoch": 5.061126648022373,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 12668
    },
    {
      "epoch": 5.061526168597683,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 12669
    },
    {
      "epoch": 5.061925689172992,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 12670
    },
    {
      "epoch": 5.062325209748302,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 12671
    },
    {
      "epoch": 5.062724730323612,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 12672
    },
    {
      "epoch": 5.063124250898921,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 12673
    },
    {
      "epoch": 5.063523771474231,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 12674
    },
    {
      "epoch": 5.063923292049541,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 12675
    },
    {
      "epoch": 5.06432281262485,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 12676
    },
    {
      "epoch": 5.06472233320016,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 12677
    },
    {
      "epoch": 5.065121853775469,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 12678
    },
    {
      "epoch": 5.065521374350779,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 12679
    },
    {
      "epoch": 5.065920894926089,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 12680
    },
    {
      "epoch": 5.066320415501398,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 12681
    },
    {
      "epoch": 5.066719936076708,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 12682
    },
    {
      "epoch": 5.0671194566520175,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 12683
    },
    {
      "epoch": 5.067518977227327,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 12684
    },
    {
      "epoch": 5.067918497802637,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 12685
    },
    {
      "epoch": 5.068318018377947,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 12686
    },
    {
      "epoch": 5.068717538953256,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 12687
    },
    {
      "epoch": 5.0691170595285655,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 12688
    },
    {
      "epoch": 5.069516580103875,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 12689
    },
    {
      "epoch": 5.069916100679185,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 12690
    },
    {
      "epoch": 5.070315621254495,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 12691
    },
    {
      "epoch": 5.070715141829805,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 12692
    },
    {
      "epoch": 5.0711146624051135,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 12693
    },
    {
      "epoch": 5.071514182980423,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 12694
    },
    {
      "epoch": 5.071913703555733,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 12695
    },
    {
      "epoch": 5.072313224131043,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 12696
    },
    {
      "epoch": 5.072712744706353,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 12697
    },
    {
      "epoch": 5.073112265281662,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 12698
    },
    {
      "epoch": 5.073511785856971,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 12699
    },
    {
      "epoch": 5.073911306432281,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 12700
    },
    {
      "epoch": 5.074310827007591,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 12701
    },
    {
      "epoch": 5.074710347582901,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0615,
      "step": 12702
    },
    {
      "epoch": 5.0751098681582105,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 12703
    },
    {
      "epoch": 5.075509388733519,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 12704
    },
    {
      "epoch": 5.075908909308829,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 12705
    },
    {
      "epoch": 5.076308429884139,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 12706
    },
    {
      "epoch": 5.076707950459449,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 12707
    },
    {
      "epoch": 5.077107471034759,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 12708
    },
    {
      "epoch": 5.077506991610068,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 12709
    },
    {
      "epoch": 5.077906512185377,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0671,
      "step": 12710
    },
    {
      "epoch": 5.078306032760687,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 12711
    },
    {
      "epoch": 5.078705553335997,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 12712
    },
    {
      "epoch": 5.079105073911307,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 12713
    },
    {
      "epoch": 5.079504594486616,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 12714
    },
    {
      "epoch": 5.079904115061925,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 12715
    },
    {
      "epoch": 5.080303635637235,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 12716
    },
    {
      "epoch": 5.080703156212545,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 12717
    },
    {
      "epoch": 5.081102676787855,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 12718
    },
    {
      "epoch": 5.081502197363164,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 12719
    },
    {
      "epoch": 5.081901717938474,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 12720
    },
    {
      "epoch": 5.082301238513783,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 12721
    },
    {
      "epoch": 5.082700759089093,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 12722
    },
    {
      "epoch": 5.083100279664403,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 12723
    },
    {
      "epoch": 5.0834998002397125,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 12724
    },
    {
      "epoch": 5.083899320815022,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 12725
    },
    {
      "epoch": 5.084298841390331,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 12726
    },
    {
      "epoch": 5.084698361965641,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 12727
    },
    {
      "epoch": 5.085097882540951,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 12728
    },
    {
      "epoch": 5.0854974031162605,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 12729
    },
    {
      "epoch": 5.08589692369157,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 12730
    },
    {
      "epoch": 5.08629644426688,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 12731
    },
    {
      "epoch": 5.086695964842189,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 12732
    },
    {
      "epoch": 5.087095485417499,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 12733
    },
    {
      "epoch": 5.087495005992809,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 12734
    },
    {
      "epoch": 5.087894526568118,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 12735
    },
    {
      "epoch": 5.088294047143428,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 12736
    },
    {
      "epoch": 5.088693567718738,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 12737
    },
    {
      "epoch": 5.089093088294047,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 12738
    },
    {
      "epoch": 5.089492608869357,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0608,
      "step": 12739
    },
    {
      "epoch": 5.089892129444666,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 12740
    },
    {
      "epoch": 5.090291650019976,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 12741
    },
    {
      "epoch": 5.090691170595286,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 12742
    },
    {
      "epoch": 5.091090691170595,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 12743
    },
    {
      "epoch": 5.091490211745905,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 12744
    },
    {
      "epoch": 5.0918897323212144,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 12745
    },
    {
      "epoch": 5.092289252896524,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 12746
    },
    {
      "epoch": 5.092688773471834,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 12747
    },
    {
      "epoch": 5.093088294047144,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0648,
      "step": 12748
    },
    {
      "epoch": 5.093487814622453,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0632,
      "step": 12749
    },
    {
      "epoch": 5.0938873351977625,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 12750
    },
    {
      "epoch": 5.094286855773072,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 12751
    },
    {
      "epoch": 5.094686376348382,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 12752
    },
    {
      "epoch": 5.095085896923692,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 12753
    },
    {
      "epoch": 5.095485417499002,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 12754
    },
    {
      "epoch": 5.0958849380743105,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 12755
    },
    {
      "epoch": 5.09628445864962,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 12756
    },
    {
      "epoch": 5.09668397922493,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 12757
    },
    {
      "epoch": 5.09708349980024,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 12758
    },
    {
      "epoch": 5.09748302037555,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 12759
    },
    {
      "epoch": 5.097882540950859,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 12760
    },
    {
      "epoch": 5.098282061526168,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 12761
    },
    {
      "epoch": 5.098681582101478,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 12762
    },
    {
      "epoch": 5.099081102676788,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 12763
    },
    {
      "epoch": 5.099480623252098,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 12764
    },
    {
      "epoch": 5.0998801438274075,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 12765
    },
    {
      "epoch": 5.100279664402716,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 12766
    },
    {
      "epoch": 5.100679184978026,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 12767
    },
    {
      "epoch": 5.101078705553336,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 12768
    },
    {
      "epoch": 5.101478226128646,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 12769
    },
    {
      "epoch": 5.1018777467039556,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 12770
    },
    {
      "epoch": 5.1022772672792644,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 12771
    },
    {
      "epoch": 5.102676787854574,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 12772
    },
    {
      "epoch": 5.103076308429884,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 12773
    },
    {
      "epoch": 5.103475829005194,
      "grad_norm": 0.28125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 12774
    },
    {
      "epoch": 5.103875349580504,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 12775
    },
    {
      "epoch": 5.104274870155813,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 12776
    },
    {
      "epoch": 5.104674390731122,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 12777
    },
    {
      "epoch": 5.105073911306432,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0005,
      "loss": 1.0617,
      "step": 12778
    },
    {
      "epoch": 5.105473431881742,
      "grad_norm": 0.609375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 12779
    },
    {
      "epoch": 5.105872952457052,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 12780
    },
    {
      "epoch": 5.106272473032361,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 12781
    },
    {
      "epoch": 5.106671993607671,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 12782
    },
    {
      "epoch": 5.10707151418298,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 12783
    },
    {
      "epoch": 5.10747103475829,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 12784
    },
    {
      "epoch": 5.1078705553336,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 12785
    },
    {
      "epoch": 5.1082700759089095,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 12786
    },
    {
      "epoch": 5.108669596484219,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 12787
    },
    {
      "epoch": 5.109069117059528,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 12788
    },
    {
      "epoch": 5.109468637634838,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0005,
      "loss": 1.0597,
      "step": 12789
    },
    {
      "epoch": 5.109868158210148,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 12790
    },
    {
      "epoch": 5.1102676787854575,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 12791
    },
    {
      "epoch": 5.110667199360767,
      "grad_norm": 0.421875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 12792
    },
    {
      "epoch": 5.111066719936077,
      "grad_norm": 0.625,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 12793
    },
    {
      "epoch": 5.111466240511386,
      "grad_norm": 0.515625,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 12794
    },
    {
      "epoch": 5.111865761086696,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 12795
    },
    {
      "epoch": 5.112265281662006,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0005,
      "loss": 1.0602,
      "step": 12796
    },
    {
      "epoch": 5.112664802237315,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 12797
    },
    {
      "epoch": 5.113064322812625,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 12798
    },
    {
      "epoch": 5.113463843387935,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 12799
    },
    {
      "epoch": 5.113863363963244,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 12800
    },
    {
      "epoch": 5.114262884538554,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 12801
    },
    {
      "epoch": 5.114662405113863,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 12802
    },
    {
      "epoch": 5.115061925689173,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 12803
    },
    {
      "epoch": 5.115461446264483,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 12804
    },
    {
      "epoch": 5.115860966839792,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 12805
    },
    {
      "epoch": 5.116260487415102,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 12806
    },
    {
      "epoch": 5.116660007990411,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 12807
    },
    {
      "epoch": 5.117059528565721,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 12808
    },
    {
      "epoch": 5.117459049141031,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 12809
    },
    {
      "epoch": 5.117858569716341,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 12810
    },
    {
      "epoch": 5.11825809029165,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 12811
    },
    {
      "epoch": 5.1186576108669595,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 12812
    },
    {
      "epoch": 5.119057131442269,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 12813
    },
    {
      "epoch": 5.119456652017579,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 12814
    },
    {
      "epoch": 5.119856172592889,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 12815
    },
    {
      "epoch": 5.120255693168198,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 12816
    },
    {
      "epoch": 5.1206552137435075,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 12817
    },
    {
      "epoch": 5.121054734318817,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 12818
    },
    {
      "epoch": 5.121454254894127,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 12819
    },
    {
      "epoch": 5.121853775469437,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 12820
    },
    {
      "epoch": 5.122253296044747,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 12821
    },
    {
      "epoch": 5.122652816620056,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 12822
    },
    {
      "epoch": 5.123052337195365,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 12823
    },
    {
      "epoch": 5.123451857770675,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 12824
    },
    {
      "epoch": 5.123851378345985,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 12825
    },
    {
      "epoch": 5.124250898921295,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 12826
    },
    {
      "epoch": 5.1246504194966045,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 12827
    },
    {
      "epoch": 5.125049940071913,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 12828
    },
    {
      "epoch": 5.125449460647223,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 12829
    },
    {
      "epoch": 5.125848981222533,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 12830
    },
    {
      "epoch": 5.126248501797843,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 12831
    },
    {
      "epoch": 5.1266480223731525,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 12832
    },
    {
      "epoch": 5.127047542948461,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 12833
    },
    {
      "epoch": 5.127447063523771,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 12834
    },
    {
      "epoch": 5.127846584099081,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 12835
    },
    {
      "epoch": 5.128246104674391,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 12836
    },
    {
      "epoch": 5.128645625249701,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 12837
    },
    {
      "epoch": 5.12904514582501,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 12838
    },
    {
      "epoch": 5.129444666400319,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 12839
    },
    {
      "epoch": 5.129844186975629,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 12840
    },
    {
      "epoch": 5.130243707550939,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 12841
    },
    {
      "epoch": 5.130643228126249,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 12842
    },
    {
      "epoch": 5.131042748701558,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 12843
    },
    {
      "epoch": 5.131442269276867,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 12844
    },
    {
      "epoch": 5.131841789852177,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 12845
    },
    {
      "epoch": 5.132241310427487,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 12846
    },
    {
      "epoch": 5.132640831002797,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 12847
    },
    {
      "epoch": 5.1330403515781065,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 12848
    },
    {
      "epoch": 5.133439872153416,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 12849
    },
    {
      "epoch": 5.133839392728725,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 12850
    },
    {
      "epoch": 5.134238913304035,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 12851
    },
    {
      "epoch": 5.134638433879345,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 12852
    },
    {
      "epoch": 5.1350379544546545,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 12853
    },
    {
      "epoch": 5.135437475029964,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 12854
    },
    {
      "epoch": 5.135836995605274,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 12855
    },
    {
      "epoch": 5.136236516180583,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 12856
    },
    {
      "epoch": 5.136636036755893,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 12857
    },
    {
      "epoch": 5.1370355573312025,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 12858
    },
    {
      "epoch": 5.137435077906512,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 12859
    },
    {
      "epoch": 5.137834598481822,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 12860
    },
    {
      "epoch": 5.138234119057131,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 12861
    },
    {
      "epoch": 5.138633639632441,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 12862
    },
    {
      "epoch": 5.139033160207751,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 12863
    },
    {
      "epoch": 5.13943268078306,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 12864
    },
    {
      "epoch": 5.13983220135837,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 12865
    },
    {
      "epoch": 5.14023172193368,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 12866
    },
    {
      "epoch": 5.140631242508989,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 12867
    },
    {
      "epoch": 5.141030763084299,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 12868
    },
    {
      "epoch": 5.141430283659608,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 12869
    },
    {
      "epoch": 5.141829804234918,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 12870
    },
    {
      "epoch": 5.142229324810228,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 12871
    },
    {
      "epoch": 5.142628845385538,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 12872
    },
    {
      "epoch": 5.143028365960847,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 12873
    },
    {
      "epoch": 5.1434278865361565,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 12874
    },
    {
      "epoch": 5.143827407111466,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 12875
    },
    {
      "epoch": 5.144226927686776,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 12876
    },
    {
      "epoch": 5.144626448262086,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 12877
    },
    {
      "epoch": 5.145025968837395,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 12878
    },
    {
      "epoch": 5.1454254894127045,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 12879
    },
    {
      "epoch": 5.145825009988014,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 12880
    },
    {
      "epoch": 5.146224530563324,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 12881
    },
    {
      "epoch": 5.146624051138634,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 12882
    },
    {
      "epoch": 5.147023571713944,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 12883
    },
    {
      "epoch": 5.1474230922892525,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 12884
    },
    {
      "epoch": 5.147822612864562,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 12885
    },
    {
      "epoch": 5.148222133439872,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 12886
    },
    {
      "epoch": 5.148621654015182,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 12887
    },
    {
      "epoch": 5.149021174590492,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 12888
    },
    {
      "epoch": 5.1494206951658015,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 12889
    },
    {
      "epoch": 5.14982021574111,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0643,
      "step": 12890
    },
    {
      "epoch": 5.15021973631642,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 12891
    },
    {
      "epoch": 5.15061925689173,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 12892
    },
    {
      "epoch": 5.15101877746704,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 12893
    },
    {
      "epoch": 5.1514182980423495,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 12894
    },
    {
      "epoch": 5.151817818617658,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 12895
    },
    {
      "epoch": 5.152217339192968,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 12896
    },
    {
      "epoch": 5.152616859768278,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 12897
    },
    {
      "epoch": 5.153016380343588,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 12898
    },
    {
      "epoch": 5.153415900918898,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 12899
    },
    {
      "epoch": 5.153815421494207,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 12900
    },
    {
      "epoch": 5.154214942069516,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 12901
    },
    {
      "epoch": 5.154614462644826,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 12902
    },
    {
      "epoch": 5.155013983220136,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 12903
    },
    {
      "epoch": 5.155413503795446,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 12904
    },
    {
      "epoch": 5.155813024370755,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 12905
    },
    {
      "epoch": 5.156212544946065,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 12906
    },
    {
      "epoch": 5.156612065521374,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 12907
    },
    {
      "epoch": 5.157011586096684,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 12908
    },
    {
      "epoch": 5.157411106671994,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 12909
    },
    {
      "epoch": 5.157810627247303,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 12910
    },
    {
      "epoch": 5.158210147822613,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 12911
    },
    {
      "epoch": 5.158609668397922,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 12912
    },
    {
      "epoch": 5.159009188973232,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 12913
    },
    {
      "epoch": 5.159408709548542,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 12914
    },
    {
      "epoch": 5.1598082301238515,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 12915
    },
    {
      "epoch": 5.160207750699161,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 12916
    },
    {
      "epoch": 5.160607271274471,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 12917
    },
    {
      "epoch": 5.16100679184978,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 12918
    },
    {
      "epoch": 5.16140631242509,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 12919
    },
    {
      "epoch": 5.1618058330003995,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 12920
    },
    {
      "epoch": 5.162205353575709,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 12921
    },
    {
      "epoch": 5.162604874151019,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 12922
    },
    {
      "epoch": 5.163004394726328,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 12923
    },
    {
      "epoch": 5.163403915301638,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 12924
    },
    {
      "epoch": 5.163803435876948,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 12925
    },
    {
      "epoch": 5.164202956452257,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 12926
    },
    {
      "epoch": 5.164602477027567,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 12927
    },
    {
      "epoch": 5.165001997602877,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 12928
    },
    {
      "epoch": 5.165401518178186,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 12929
    },
    {
      "epoch": 5.165801038753496,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 12930
    },
    {
      "epoch": 5.166200559328805,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 12931
    },
    {
      "epoch": 5.166600079904115,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 12932
    },
    {
      "epoch": 5.166999600479425,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 12933
    },
    {
      "epoch": 5.167399121054734,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 12934
    },
    {
      "epoch": 5.167798641630044,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 12935
    },
    {
      "epoch": 5.168198162205353,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 12936
    },
    {
      "epoch": 5.168597682780663,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 12937
    },
    {
      "epoch": 5.168997203355973,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 12938
    },
    {
      "epoch": 5.169396723931283,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 12939
    },
    {
      "epoch": 5.169796244506592,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 12940
    },
    {
      "epoch": 5.1701957650819015,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 12941
    },
    {
      "epoch": 5.170595285657211,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 12942
    },
    {
      "epoch": 5.170994806232521,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 12943
    },
    {
      "epoch": 5.171394326807831,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 12944
    },
    {
      "epoch": 5.171793847383141,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 12945
    },
    {
      "epoch": 5.1721933679584495,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 12946
    },
    {
      "epoch": 5.172592888533759,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 12947
    },
    {
      "epoch": 5.172992409109069,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 12948
    },
    {
      "epoch": 5.173391929684379,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 12949
    },
    {
      "epoch": 5.173791450259689,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 12950
    },
    {
      "epoch": 5.174190970834998,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 12951
    },
    {
      "epoch": 5.174590491410307,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 12952
    },
    {
      "epoch": 5.174990011985617,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 12953
    },
    {
      "epoch": 5.175389532560927,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 12954
    },
    {
      "epoch": 5.175789053136237,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 12955
    },
    {
      "epoch": 5.1761885737115465,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 12956
    },
    {
      "epoch": 5.176588094286855,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 12957
    },
    {
      "epoch": 5.176987614862165,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 12958
    },
    {
      "epoch": 5.177387135437475,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 12959
    },
    {
      "epoch": 5.177786656012785,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 12960
    },
    {
      "epoch": 5.1781861765880945,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0612,
      "step": 12961
    },
    {
      "epoch": 5.178585697163404,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 12962
    },
    {
      "epoch": 5.178985217738713,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 12963
    },
    {
      "epoch": 5.179384738314023,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 12964
    },
    {
      "epoch": 5.179784258889333,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 12965
    },
    {
      "epoch": 5.180183779464643,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 12966
    },
    {
      "epoch": 5.180583300039952,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 12967
    },
    {
      "epoch": 5.180982820615261,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 12968
    },
    {
      "epoch": 5.181382341190571,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 12969
    },
    {
      "epoch": 5.181781861765881,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 12970
    },
    {
      "epoch": 5.182181382341191,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 12971
    },
    {
      "epoch": 5.1825809029165,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 12972
    },
    {
      "epoch": 5.18298042349181,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 12973
    },
    {
      "epoch": 5.183379944067119,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 12974
    },
    {
      "epoch": 5.183779464642429,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 12975
    },
    {
      "epoch": 5.184178985217739,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 12976
    },
    {
      "epoch": 5.1845785057930485,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 12977
    },
    {
      "epoch": 5.184978026368358,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 12978
    },
    {
      "epoch": 5.185377546943668,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0635,
      "step": 12979
    },
    {
      "epoch": 5.185777067518977,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 12980
    },
    {
      "epoch": 5.186176588094287,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 12981
    },
    {
      "epoch": 5.1865761086695965,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 12982
    },
    {
      "epoch": 5.186975629244906,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 12983
    },
    {
      "epoch": 5.187375149820216,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 12984
    },
    {
      "epoch": 5.187774670395525,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 12985
    },
    {
      "epoch": 5.188174190970835,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 12986
    },
    {
      "epoch": 5.1885737115461446,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 12987
    },
    {
      "epoch": 5.188973232121454,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 12988
    },
    {
      "epoch": 5.189372752696764,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 12989
    },
    {
      "epoch": 5.189772273272074,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 12990
    },
    {
      "epoch": 5.190171793847383,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 12991
    },
    {
      "epoch": 5.190571314422693,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 12992
    },
    {
      "epoch": 5.190970834998002,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 12993
    },
    {
      "epoch": 5.191370355573312,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 12994
    },
    {
      "epoch": 5.191769876148622,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 12995
    },
    {
      "epoch": 5.192169396723932,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 12996
    },
    {
      "epoch": 5.192568917299241,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 12997
    },
    {
      "epoch": 5.19296843787455,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 12998
    },
    {
      "epoch": 5.19336795844986,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 12999
    },
    {
      "epoch": 5.19376747902517,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 13000
    },
    {
      "epoch": 5.19416699960048,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 13001
    },
    {
      "epoch": 5.194566520175789,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 13002
    },
    {
      "epoch": 5.1949660407510985,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 13003
    },
    {
      "epoch": 5.195365561326408,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 13004
    },
    {
      "epoch": 5.195765081901718,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 13005
    },
    {
      "epoch": 5.196164602477028,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 13006
    },
    {
      "epoch": 5.196564123052338,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 13007
    },
    {
      "epoch": 5.1969636436276465,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 13008
    },
    {
      "epoch": 5.197363164202956,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 13009
    },
    {
      "epoch": 5.197762684778266,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 13010
    },
    {
      "epoch": 5.198162205353576,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 13011
    },
    {
      "epoch": 5.198561725928886,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 13012
    },
    {
      "epoch": 5.1989612465041946,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 13013
    },
    {
      "epoch": 5.199360767079504,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 13014
    },
    {
      "epoch": 5.199760287654814,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 13015
    },
    {
      "epoch": 5.200159808230124,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 13016
    },
    {
      "epoch": 5.200559328805434,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 13017
    },
    {
      "epoch": 5.2009588493807435,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 13018
    },
    {
      "epoch": 5.201358369956052,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 13019
    },
    {
      "epoch": 5.201757890531362,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 13020
    },
    {
      "epoch": 5.202157411106672,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 13021
    },
    {
      "epoch": 5.202556931681982,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 13022
    },
    {
      "epoch": 5.2029564522572915,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 13023
    },
    {
      "epoch": 5.2033559728326,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 13024
    },
    {
      "epoch": 5.20375549340791,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 13025
    },
    {
      "epoch": 5.20415501398322,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 13026
    },
    {
      "epoch": 5.20455453455853,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 13027
    },
    {
      "epoch": 5.20495405513384,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 13028
    },
    {
      "epoch": 5.205353575709149,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 13029
    },
    {
      "epoch": 5.205753096284458,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 13030
    },
    {
      "epoch": 5.206152616859768,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 13031
    },
    {
      "epoch": 5.206552137435078,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 13032
    },
    {
      "epoch": 5.206951658010388,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 13033
    },
    {
      "epoch": 5.207351178585697,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 13034
    },
    {
      "epoch": 5.207750699161007,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 13035
    },
    {
      "epoch": 5.208150219736316,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 13036
    },
    {
      "epoch": 5.208549740311626,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 13037
    },
    {
      "epoch": 5.208949260886936,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 13038
    },
    {
      "epoch": 5.2093487814622454,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 13039
    },
    {
      "epoch": 5.209748302037555,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 13040
    },
    {
      "epoch": 5.210147822612864,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 13041
    },
    {
      "epoch": 5.210547343188174,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 13042
    },
    {
      "epoch": 5.210946863763484,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 13043
    },
    {
      "epoch": 5.2113463843387935,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 13044
    },
    {
      "epoch": 5.211745904914103,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 13045
    },
    {
      "epoch": 5.212145425489413,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 13046
    },
    {
      "epoch": 5.212544946064722,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 13047
    },
    {
      "epoch": 5.212944466640032,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 13048
    },
    {
      "epoch": 5.2133439872153415,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 13049
    },
    {
      "epoch": 5.213743507790651,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 13050
    },
    {
      "epoch": 5.214143028365961,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 13051
    },
    {
      "epoch": 5.214542548941271,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 13052
    },
    {
      "epoch": 5.21494206951658,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 13053
    },
    {
      "epoch": 5.21534159009189,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 13054
    },
    {
      "epoch": 5.215741110667199,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 13055
    },
    {
      "epoch": 5.216140631242509,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 13056
    },
    {
      "epoch": 5.216540151817819,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 13057
    },
    {
      "epoch": 5.216939672393128,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 13058
    },
    {
      "epoch": 5.217339192968438,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 13059
    },
    {
      "epoch": 5.217738713543747,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 13060
    },
    {
      "epoch": 5.218138234119057,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 13061
    },
    {
      "epoch": 5.218537754694367,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 13062
    },
    {
      "epoch": 5.218937275269677,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 13063
    },
    {
      "epoch": 5.219336795844986,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 13064
    },
    {
      "epoch": 5.2197363164202955,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 13065
    },
    {
      "epoch": 5.220135836995605,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 13066
    },
    {
      "epoch": 5.220535357570915,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 13067
    },
    {
      "epoch": 5.220934878146225,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 13068
    },
    {
      "epoch": 5.221334398721535,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 13069
    },
    {
      "epoch": 5.2217339192968435,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 13070
    },
    {
      "epoch": 5.222133439872153,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 13071
    },
    {
      "epoch": 5.222532960447463,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 13072
    },
    {
      "epoch": 5.222932481022773,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 13073
    },
    {
      "epoch": 5.223332001598083,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 13074
    },
    {
      "epoch": 5.2237315221733915,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 13075
    },
    {
      "epoch": 5.224131042748701,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 13076
    },
    {
      "epoch": 5.224530563324011,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 13077
    },
    {
      "epoch": 5.224930083899321,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 13078
    },
    {
      "epoch": 5.225329604474631,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 13079
    },
    {
      "epoch": 5.2257291250499405,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 13080
    },
    {
      "epoch": 5.226128645625249,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 13081
    },
    {
      "epoch": 5.226528166200559,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 13082
    },
    {
      "epoch": 5.226927686775869,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 13083
    },
    {
      "epoch": 5.227327207351179,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 13084
    },
    {
      "epoch": 5.2277267279264885,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 13085
    },
    {
      "epoch": 5.228126248501797,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 13086
    },
    {
      "epoch": 5.228525769077107,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 13087
    },
    {
      "epoch": 5.228925289652417,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 13088
    },
    {
      "epoch": 5.229324810227727,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 13089
    },
    {
      "epoch": 5.229724330803037,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 13090
    },
    {
      "epoch": 5.230123851378346,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 13091
    },
    {
      "epoch": 5.230523371953655,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 13092
    },
    {
      "epoch": 5.230922892528965,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 13093
    },
    {
      "epoch": 5.231322413104275,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 13094
    },
    {
      "epoch": 5.231721933679585,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 13095
    },
    {
      "epoch": 5.232121454254894,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 13096
    },
    {
      "epoch": 5.232520974830204,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 13097
    },
    {
      "epoch": 5.232920495405513,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 13098
    },
    {
      "epoch": 5.233320015980823,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 13099
    },
    {
      "epoch": 5.233719536556133,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 13100
    },
    {
      "epoch": 5.234119057131442,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 13101
    },
    {
      "epoch": 5.234518577706752,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 13102
    },
    {
      "epoch": 5.234918098282061,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 13103
    },
    {
      "epoch": 5.235317618857371,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 13104
    },
    {
      "epoch": 5.235717139432681,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 13105
    },
    {
      "epoch": 5.2361166600079905,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 13106
    },
    {
      "epoch": 5.2365161805833,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 13107
    },
    {
      "epoch": 5.23691570115861,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 13108
    },
    {
      "epoch": 5.237315221733919,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 13109
    },
    {
      "epoch": 5.237714742309229,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 13110
    },
    {
      "epoch": 5.2381142628845385,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 13111
    },
    {
      "epoch": 5.238513783459848,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 13112
    },
    {
      "epoch": 5.238913304035158,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 13113
    },
    {
      "epoch": 5.239312824610467,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 13114
    },
    {
      "epoch": 5.239712345185777,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 13115
    },
    {
      "epoch": 5.240111865761087,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 13116
    },
    {
      "epoch": 5.240511386336396,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 13117
    },
    {
      "epoch": 5.240910906911706,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 13118
    },
    {
      "epoch": 5.241310427487016,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 13119
    },
    {
      "epoch": 5.241709948062325,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 13120
    },
    {
      "epoch": 5.242109468637635,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 13121
    },
    {
      "epoch": 5.242508989212944,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 13122
    },
    {
      "epoch": 5.242908509788254,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0586,
      "step": 13123
    },
    {
      "epoch": 5.243308030363564,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 13124
    },
    {
      "epoch": 5.243707550938874,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 13125
    },
    {
      "epoch": 5.244107071514183,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 13126
    },
    {
      "epoch": 5.244506592089492,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 13127
    },
    {
      "epoch": 5.244906112664802,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 13128
    },
    {
      "epoch": 5.245305633240112,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 13129
    },
    {
      "epoch": 5.245705153815422,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 13130
    },
    {
      "epoch": 5.246104674390731,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 13131
    },
    {
      "epoch": 5.2465041949660405,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 13132
    },
    {
      "epoch": 5.24690371554135,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 13133
    },
    {
      "epoch": 5.24730323611666,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 13134
    },
    {
      "epoch": 5.24770275669197,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 13135
    },
    {
      "epoch": 5.24810227726728,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 13136
    },
    {
      "epoch": 5.2485017978425885,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 13137
    },
    {
      "epoch": 5.248901318417898,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 13138
    },
    {
      "epoch": 5.249300838993208,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 13139
    },
    {
      "epoch": 5.249700359568518,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 13140
    },
    {
      "epoch": 5.250099880143828,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 13141
    },
    {
      "epoch": 5.2504994007191375,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 13142
    },
    {
      "epoch": 5.250898921294446,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 13143
    },
    {
      "epoch": 5.251298441869756,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 13144
    },
    {
      "epoch": 5.251697962445066,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 13145
    },
    {
      "epoch": 5.252097483020376,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 13146
    },
    {
      "epoch": 5.2524970035956855,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 13147
    },
    {
      "epoch": 5.252896524170994,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 13148
    },
    {
      "epoch": 5.253296044746304,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 13149
    },
    {
      "epoch": 5.253695565321614,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 13150
    },
    {
      "epoch": 5.254095085896924,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 13151
    },
    {
      "epoch": 5.2544946064722335,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 13152
    },
    {
      "epoch": 5.254894127047543,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 13153
    },
    {
      "epoch": 5.255293647622852,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 13154
    },
    {
      "epoch": 5.255693168198162,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 13155
    },
    {
      "epoch": 5.256092688773472,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 13156
    },
    {
      "epoch": 5.256492209348782,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 13157
    },
    {
      "epoch": 5.256891729924091,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 13158
    },
    {
      "epoch": 5.257291250499401,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 13159
    },
    {
      "epoch": 5.25769077107471,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 13160
    },
    {
      "epoch": 5.25809029165002,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 13161
    },
    {
      "epoch": 5.25848981222533,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 13162
    },
    {
      "epoch": 5.258889332800639,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 13163
    },
    {
      "epoch": 5.259288853375949,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 13164
    },
    {
      "epoch": 5.259688373951258,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 13165
    },
    {
      "epoch": 5.260087894526568,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 13166
    },
    {
      "epoch": 5.260487415101878,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 13167
    },
    {
      "epoch": 5.2608869356771875,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 13168
    },
    {
      "epoch": 5.261286456252497,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 13169
    },
    {
      "epoch": 5.261685976827807,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 13170
    },
    {
      "epoch": 5.262085497403116,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 13171
    },
    {
      "epoch": 5.262485017978426,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 13172
    },
    {
      "epoch": 5.2628845385537355,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 13173
    },
    {
      "epoch": 5.263284059129045,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 13174
    },
    {
      "epoch": 5.263683579704355,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 13175
    },
    {
      "epoch": 5.264083100279665,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 13176
    },
    {
      "epoch": 5.264482620854974,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 13177
    },
    {
      "epoch": 5.2648821414302835,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 13178
    },
    {
      "epoch": 5.265281662005593,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 13179
    },
    {
      "epoch": 5.265681182580903,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 13180
    },
    {
      "epoch": 5.266080703156213,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 13181
    },
    {
      "epoch": 5.266480223731522,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 13182
    },
    {
      "epoch": 5.266879744306832,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 13183
    },
    {
      "epoch": 5.267279264882141,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 13184
    },
    {
      "epoch": 5.267678785457451,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 13185
    },
    {
      "epoch": 5.268078306032761,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 13186
    },
    {
      "epoch": 5.26847782660807,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 13187
    },
    {
      "epoch": 5.26887734718338,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 13188
    },
    {
      "epoch": 5.269276867758689,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 13189
    },
    {
      "epoch": 5.269676388333999,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 13190
    },
    {
      "epoch": 5.270075908909309,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 13191
    },
    {
      "epoch": 5.270475429484619,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 13192
    },
    {
      "epoch": 5.270874950059928,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 13193
    },
    {
      "epoch": 5.2712744706352375,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 13194
    },
    {
      "epoch": 5.271673991210547,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 13195
    },
    {
      "epoch": 5.272073511785857,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 13196
    },
    {
      "epoch": 5.272473032361167,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 13197
    },
    {
      "epoch": 5.272872552936477,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 13198
    },
    {
      "epoch": 5.2732720735117855,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 13199
    },
    {
      "epoch": 5.273671594087095,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 13200
    },
    {
      "epoch": 5.274071114662405,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 13201
    },
    {
      "epoch": 5.274470635237715,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 13202
    },
    {
      "epoch": 5.274870155813025,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 13203
    },
    {
      "epoch": 5.2752696763883336,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 13204
    },
    {
      "epoch": 5.275669196963643,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 13205
    },
    {
      "epoch": 5.276068717538953,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 13206
    },
    {
      "epoch": 5.276468238114263,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 13207
    },
    {
      "epoch": 5.276867758689573,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 13208
    },
    {
      "epoch": 5.2772672792648825,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 13209
    },
    {
      "epoch": 5.277666799840191,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 13210
    },
    {
      "epoch": 5.278066320415501,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 13211
    },
    {
      "epoch": 5.278465840990811,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 13212
    },
    {
      "epoch": 5.278865361566121,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 13213
    },
    {
      "epoch": 5.2792648821414305,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 13214
    },
    {
      "epoch": 5.27966440271674,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 13215
    },
    {
      "epoch": 5.280063923292049,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 13216
    },
    {
      "epoch": 5.280463443867359,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 13217
    },
    {
      "epoch": 5.280862964442669,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 13218
    },
    {
      "epoch": 5.281262485017979,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 13219
    },
    {
      "epoch": 5.281662005593288,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 13220
    },
    {
      "epoch": 5.282061526168597,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 13221
    },
    {
      "epoch": 5.282461046743907,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 13222
    },
    {
      "epoch": 5.282860567319217,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 13223
    },
    {
      "epoch": 5.283260087894527,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 13224
    },
    {
      "epoch": 5.283659608469836,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 13225
    },
    {
      "epoch": 5.284059129045146,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 13226
    },
    {
      "epoch": 5.284458649620455,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 13227
    },
    {
      "epoch": 5.284858170195765,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 13228
    },
    {
      "epoch": 5.285257690771075,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 13229
    },
    {
      "epoch": 5.285657211346384,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 13230
    },
    {
      "epoch": 5.286056731921694,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 13231
    },
    {
      "epoch": 5.286456252497004,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 13232
    },
    {
      "epoch": 5.286855773072313,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 13233
    },
    {
      "epoch": 5.287255293647623,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 13234
    },
    {
      "epoch": 5.2876548142229325,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 13235
    },
    {
      "epoch": 5.288054334798242,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 13236
    },
    {
      "epoch": 5.288453855373552,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 13237
    },
    {
      "epoch": 5.288853375948861,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 13238
    },
    {
      "epoch": 5.289252896524171,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 13239
    },
    {
      "epoch": 5.2896524170994805,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 13240
    },
    {
      "epoch": 5.29005193767479,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 13241
    },
    {
      "epoch": 5.2904514582501,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 13242
    },
    {
      "epoch": 5.29085097882541,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 13243
    },
    {
      "epoch": 5.291250499400719,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 13244
    },
    {
      "epoch": 5.291650019976029,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 13245
    },
    {
      "epoch": 5.292049540551338,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 13246
    },
    {
      "epoch": 5.292449061126648,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 13247
    },
    {
      "epoch": 5.292848581701958,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 13248
    },
    {
      "epoch": 5.293248102277268,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 13249
    },
    {
      "epoch": 5.293647622852577,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 13250
    },
    {
      "epoch": 5.294047143427886,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 13251
    },
    {
      "epoch": 5.294446664003196,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 13252
    },
    {
      "epoch": 5.294846184578506,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 13253
    },
    {
      "epoch": 5.295245705153816,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 13254
    },
    {
      "epoch": 5.295645225729125,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 13255
    },
    {
      "epoch": 5.2960447463044344,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 13256
    },
    {
      "epoch": 5.296444266879744,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 13257
    },
    {
      "epoch": 5.296843787455054,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 13258
    },
    {
      "epoch": 5.297243308030364,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 13259
    },
    {
      "epoch": 5.297642828605674,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 13260
    },
    {
      "epoch": 5.2980423491809825,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 13261
    },
    {
      "epoch": 5.298441869756292,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 13262
    },
    {
      "epoch": 5.298841390331602,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 13263
    },
    {
      "epoch": 5.299240910906912,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 13264
    },
    {
      "epoch": 5.299640431482222,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 13265
    },
    {
      "epoch": 5.300039952057531,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 13266
    },
    {
      "epoch": 5.30043947263284,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 13267
    },
    {
      "epoch": 5.30083899320815,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 13268
    },
    {
      "epoch": 5.30123851378346,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 13269
    },
    {
      "epoch": 5.30163803435877,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 13270
    },
    {
      "epoch": 5.3020375549340795,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 13271
    },
    {
      "epoch": 5.302437075509388,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 13272
    },
    {
      "epoch": 5.302836596084698,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 13273
    },
    {
      "epoch": 5.303236116660008,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 13274
    },
    {
      "epoch": 5.303635637235318,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 13275
    },
    {
      "epoch": 5.3040351578106275,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 13276
    },
    {
      "epoch": 5.304434678385936,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 13277
    },
    {
      "epoch": 5.304834198961246,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 13278
    },
    {
      "epoch": 5.305233719536556,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 13279
    },
    {
      "epoch": 5.305633240111866,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 13280
    },
    {
      "epoch": 5.3060327606871756,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 13281
    },
    {
      "epoch": 5.306432281262485,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 13282
    },
    {
      "epoch": 5.306831801837794,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 13283
    },
    {
      "epoch": 5.307231322413104,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 13284
    },
    {
      "epoch": 5.307630842988414,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 13285
    },
    {
      "epoch": 5.308030363563724,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0614,
      "step": 13286
    },
    {
      "epoch": 5.308429884139033,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 13287
    },
    {
      "epoch": 5.308829404714343,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 13288
    },
    {
      "epoch": 5.309228925289652,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 13289
    },
    {
      "epoch": 5.309628445864962,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 13290
    },
    {
      "epoch": 5.310027966440272,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 13291
    },
    {
      "epoch": 5.310427487015581,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 13292
    },
    {
      "epoch": 5.310827007590891,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 13293
    },
    {
      "epoch": 5.3112265281662,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 13294
    },
    {
      "epoch": 5.31162604874151,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 13295
    },
    {
      "epoch": 5.31202556931682,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 13296
    },
    {
      "epoch": 5.3124250898921295,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 13297
    },
    {
      "epoch": 5.312824610467439,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 13298
    },
    {
      "epoch": 5.313224131042749,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 13299
    },
    {
      "epoch": 5.313623651618058,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 13300
    },
    {
      "epoch": 5.314023172193368,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 13301
    },
    {
      "epoch": 5.3144226927686775,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 13302
    },
    {
      "epoch": 5.314822213343987,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 13303
    },
    {
      "epoch": 5.315221733919297,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 13304
    },
    {
      "epoch": 5.315621254494607,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 13305
    },
    {
      "epoch": 5.316020775069916,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 13306
    },
    {
      "epoch": 5.316420295645226,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 13307
    },
    {
      "epoch": 5.316819816220535,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 13308
    },
    {
      "epoch": 5.317219336795845,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 13309
    },
    {
      "epoch": 5.317618857371155,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 13310
    },
    {
      "epoch": 5.318018377946464,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 13311
    },
    {
      "epoch": 5.318417898521774,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 13312
    },
    {
      "epoch": 5.318817419097083,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 13313
    },
    {
      "epoch": 5.319216939672393,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 13314
    },
    {
      "epoch": 5.319616460247703,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 13315
    },
    {
      "epoch": 5.320015980823013,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 13316
    },
    {
      "epoch": 5.320415501398322,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 13317
    },
    {
      "epoch": 5.320815021973631,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 13318
    },
    {
      "epoch": 5.321214542548941,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 13319
    },
    {
      "epoch": 5.321614063124251,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 13320
    },
    {
      "epoch": 5.322013583699561,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 13321
    },
    {
      "epoch": 5.322413104274871,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 13322
    },
    {
      "epoch": 5.3228126248501795,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 13323
    },
    {
      "epoch": 5.323212145425489,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 13324
    },
    {
      "epoch": 5.323611666000799,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 13325
    },
    {
      "epoch": 5.324011186576109,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 13326
    },
    {
      "epoch": 5.324410707151419,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 13327
    },
    {
      "epoch": 5.3248102277267275,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 13328
    },
    {
      "epoch": 5.325209748302037,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 13329
    },
    {
      "epoch": 5.325609268877347,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 13330
    },
    {
      "epoch": 5.326008789452657,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 13331
    },
    {
      "epoch": 5.326408310027967,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 13332
    },
    {
      "epoch": 5.3268078306032765,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0641,
      "step": 13333
    },
    {
      "epoch": 5.327207351178585,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 13334
    },
    {
      "epoch": 5.327606871753895,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 13335
    },
    {
      "epoch": 5.328006392329205,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 13336
    },
    {
      "epoch": 5.328405912904515,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 13337
    },
    {
      "epoch": 5.3288054334798245,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 13338
    },
    {
      "epoch": 5.329204954055134,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 13339
    },
    {
      "epoch": 5.329604474630443,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 13340
    },
    {
      "epoch": 5.330003995205753,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 13341
    },
    {
      "epoch": 5.330403515781063,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 13342
    },
    {
      "epoch": 5.3308030363563725,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 13343
    },
    {
      "epoch": 5.331202556931682,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 13344
    },
    {
      "epoch": 5.331602077506991,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 13345
    },
    {
      "epoch": 5.332001598082301,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 13346
    },
    {
      "epoch": 5.332401118657611,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 13347
    },
    {
      "epoch": 5.332800639232921,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 13348
    },
    {
      "epoch": 5.33320015980823,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 13349
    },
    {
      "epoch": 5.33359968038354,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 13350
    },
    {
      "epoch": 5.333999200958849,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 13351
    },
    {
      "epoch": 5.334398721534159,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 13352
    },
    {
      "epoch": 5.334798242109469,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 13353
    },
    {
      "epoch": 5.335197762684778,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 13354
    },
    {
      "epoch": 5.335597283260088,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 13355
    },
    {
      "epoch": 5.335996803835398,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 13356
    },
    {
      "epoch": 5.336396324410707,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 13357
    },
    {
      "epoch": 5.336795844986017,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 13358
    },
    {
      "epoch": 5.3371953655613265,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 13359
    },
    {
      "epoch": 5.337594886136636,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 13360
    },
    {
      "epoch": 5.337994406711946,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 13361
    },
    {
      "epoch": 5.338393927287255,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 13362
    },
    {
      "epoch": 5.338793447862565,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 13363
    },
    {
      "epoch": 5.3391929684378745,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 13364
    },
    {
      "epoch": 5.339592489013184,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 13365
    },
    {
      "epoch": 5.339992009588494,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 13366
    },
    {
      "epoch": 5.340391530163803,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 13367
    },
    {
      "epoch": 5.340791050739113,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 13368
    },
    {
      "epoch": 5.3411905713144225,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 13369
    },
    {
      "epoch": 5.341590091889732,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 13370
    },
    {
      "epoch": 5.341989612465042,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 13371
    },
    {
      "epoch": 5.342389133040352,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 13372
    },
    {
      "epoch": 5.342788653615661,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 13373
    },
    {
      "epoch": 5.343188174190971,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 13374
    },
    {
      "epoch": 5.34358769476628,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 13375
    },
    {
      "epoch": 5.34398721534159,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 13376
    },
    {
      "epoch": 5.3443867359169,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 13377
    },
    {
      "epoch": 5.34478625649221,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 13378
    },
    {
      "epoch": 5.345185777067519,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 13379
    },
    {
      "epoch": 5.345585297642828,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 13380
    },
    {
      "epoch": 5.345984818218138,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 13381
    },
    {
      "epoch": 5.346384338793448,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 13382
    },
    {
      "epoch": 5.346783859368758,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 13383
    },
    {
      "epoch": 5.347183379944067,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 13384
    },
    {
      "epoch": 5.3475829005193765,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 13385
    },
    {
      "epoch": 5.347982421094686,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 13386
    },
    {
      "epoch": 5.348381941669996,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 13387
    },
    {
      "epoch": 5.348781462245306,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 13388
    },
    {
      "epoch": 5.349180982820616,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 13389
    },
    {
      "epoch": 5.3495805033959245,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 13390
    },
    {
      "epoch": 5.349980023971234,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 13391
    },
    {
      "epoch": 5.350379544546544,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 13392
    },
    {
      "epoch": 5.350779065121854,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 13393
    },
    {
      "epoch": 5.351178585697164,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 13394
    },
    {
      "epoch": 5.351578106272473,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 13395
    },
    {
      "epoch": 5.351977626847782,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 13396
    },
    {
      "epoch": 5.352377147423092,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 13397
    },
    {
      "epoch": 5.352776667998402,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 13398
    },
    {
      "epoch": 5.353176188573712,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 13399
    },
    {
      "epoch": 5.3535757091490215,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 13400
    },
    {
      "epoch": 5.35397522972433,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 13401
    },
    {
      "epoch": 5.35437475029964,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 13402
    },
    {
      "epoch": 5.35477427087495,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 13403
    },
    {
      "epoch": 5.35517379145026,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 13404
    },
    {
      "epoch": 5.3555733120255695,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 13405
    },
    {
      "epoch": 5.355972832600879,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 13406
    },
    {
      "epoch": 5.356372353176188,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0616,
      "step": 13407
    },
    {
      "epoch": 5.356771873751498,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 13408
    },
    {
      "epoch": 5.357171394326808,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 13409
    },
    {
      "epoch": 5.357570914902118,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 13410
    },
    {
      "epoch": 5.357970435477427,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 13411
    },
    {
      "epoch": 5.358369956052737,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 13412
    },
    {
      "epoch": 5.358769476628046,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 13413
    },
    {
      "epoch": 5.359168997203356,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 13414
    },
    {
      "epoch": 5.359568517778666,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 13415
    },
    {
      "epoch": 5.359968038353975,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 13416
    },
    {
      "epoch": 5.360367558929285,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 13417
    },
    {
      "epoch": 5.360767079504594,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 13418
    },
    {
      "epoch": 5.361166600079904,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 13419
    },
    {
      "epoch": 5.361566120655214,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 13420
    },
    {
      "epoch": 5.361965641230523,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 13421
    },
    {
      "epoch": 5.362365161805833,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 13422
    },
    {
      "epoch": 5.362764682381143,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 13423
    },
    {
      "epoch": 5.363164202956452,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 13424
    },
    {
      "epoch": 5.363563723531762,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 13425
    },
    {
      "epoch": 5.3639632441070715,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 13426
    },
    {
      "epoch": 5.364362764682381,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 13427
    },
    {
      "epoch": 5.364762285257691,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 13428
    },
    {
      "epoch": 5.365161805833001,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 13429
    },
    {
      "epoch": 5.36556132640831,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 13430
    },
    {
      "epoch": 5.3659608469836195,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 13431
    },
    {
      "epoch": 5.366360367558929,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 13432
    },
    {
      "epoch": 5.366759888134239,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 13433
    },
    {
      "epoch": 5.367159408709549,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 13434
    },
    {
      "epoch": 5.367558929284858,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 13435
    },
    {
      "epoch": 5.367958449860168,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 13436
    },
    {
      "epoch": 5.368357970435477,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 13437
    },
    {
      "epoch": 5.368757491010787,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 13438
    },
    {
      "epoch": 5.369157011586097,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 13439
    },
    {
      "epoch": 5.369556532161407,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 13440
    },
    {
      "epoch": 5.369956052736716,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 13441
    },
    {
      "epoch": 5.370355573312025,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 13442
    },
    {
      "epoch": 5.370755093887335,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 13443
    },
    {
      "epoch": 5.371154614462645,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 13444
    },
    {
      "epoch": 5.371554135037955,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 13445
    },
    {
      "epoch": 5.3719536556132645,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 13446
    },
    {
      "epoch": 5.372353176188573,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 13447
    },
    {
      "epoch": 5.372752696763883,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 13448
    },
    {
      "epoch": 5.373152217339193,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 13449
    },
    {
      "epoch": 5.373551737914503,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 13450
    },
    {
      "epoch": 5.373951258489813,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 13451
    },
    {
      "epoch": 5.3743507790651215,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 13452
    },
    {
      "epoch": 5.374750299640431,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 13453
    },
    {
      "epoch": 5.375149820215741,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 13454
    },
    {
      "epoch": 5.375549340791051,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 13455
    },
    {
      "epoch": 5.375948861366361,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 13456
    },
    {
      "epoch": 5.3763483819416695,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 13457
    },
    {
      "epoch": 5.376747902516979,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 13458
    },
    {
      "epoch": 5.377147423092289,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 13459
    },
    {
      "epoch": 5.377546943667599,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 13460
    },
    {
      "epoch": 5.377946464242909,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 13461
    },
    {
      "epoch": 5.3783459848182185,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 13462
    },
    {
      "epoch": 5.378745505393527,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 13463
    },
    {
      "epoch": 5.379145025968837,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 13464
    },
    {
      "epoch": 5.379544546544147,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 13465
    },
    {
      "epoch": 5.379944067119457,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 13466
    },
    {
      "epoch": 5.3803435876947665,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 13467
    },
    {
      "epoch": 5.380743108270076,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 13468
    },
    {
      "epoch": 5.381142628845385,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 13469
    },
    {
      "epoch": 5.381542149420695,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 13470
    },
    {
      "epoch": 5.381941669996005,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 13471
    },
    {
      "epoch": 5.3823411905713145,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 13472
    },
    {
      "epoch": 5.382740711146624,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 13473
    },
    {
      "epoch": 5.383140231721933,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 13474
    },
    {
      "epoch": 5.383539752297243,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 13475
    },
    {
      "epoch": 5.383939272872553,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 13476
    },
    {
      "epoch": 5.384338793447863,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 13477
    },
    {
      "epoch": 5.384738314023172,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 13478
    },
    {
      "epoch": 5.385137834598482,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 13479
    },
    {
      "epoch": 5.385537355173791,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 13480
    },
    {
      "epoch": 5.385936875749101,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 13481
    },
    {
      "epoch": 5.386336396324411,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 13482
    },
    {
      "epoch": 5.38673591689972,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 13483
    },
    {
      "epoch": 5.38713543747503,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 13484
    },
    {
      "epoch": 5.38753495805034,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 13485
    },
    {
      "epoch": 5.387934478625649,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 13486
    },
    {
      "epoch": 5.388333999200959,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 13487
    },
    {
      "epoch": 5.3887335197762685,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 13488
    },
    {
      "epoch": 5.389133040351578,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 13489
    },
    {
      "epoch": 5.389532560926888,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 13490
    },
    {
      "epoch": 5.389932081502197,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 13491
    },
    {
      "epoch": 5.390331602077507,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 13492
    },
    {
      "epoch": 5.3907311226528165,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 13493
    },
    {
      "epoch": 5.391130643228126,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 13494
    },
    {
      "epoch": 5.391530163803436,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 13495
    },
    {
      "epoch": 5.391929684378746,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 13496
    },
    {
      "epoch": 5.392329204954055,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 13497
    },
    {
      "epoch": 5.3927287255293646,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 13498
    },
    {
      "epoch": 5.393128246104674,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 13499
    },
    {
      "epoch": 5.393527766679984,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 13500
    },
    {
      "epoch": 5.393927287255294,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 13501
    },
    {
      "epoch": 5.394326807830604,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 13502
    },
    {
      "epoch": 5.394726328405913,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 13503
    },
    {
      "epoch": 5.395125848981222,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 13504
    },
    {
      "epoch": 5.395525369556532,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 13505
    },
    {
      "epoch": 5.395924890131842,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 13506
    },
    {
      "epoch": 5.396324410707152,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 13507
    },
    {
      "epoch": 5.396723931282461,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 13508
    },
    {
      "epoch": 5.39712345185777,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 13509
    },
    {
      "epoch": 5.39752297243308,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 13510
    },
    {
      "epoch": 5.39792249300839,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 13511
    },
    {
      "epoch": 5.3983220135837,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 13512
    },
    {
      "epoch": 5.39872153415901,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 13513
    },
    {
      "epoch": 5.3991210547343185,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 13514
    },
    {
      "epoch": 5.399520575309628,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 13515
    },
    {
      "epoch": 5.399920095884938,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 13516
    },
    {
      "epoch": 5.400319616460248,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 13517
    },
    {
      "epoch": 5.400719137035558,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 13518
    },
    {
      "epoch": 5.401118657610867,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 13519
    },
    {
      "epoch": 5.401518178186176,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 13520
    },
    {
      "epoch": 5.401917698761486,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 13521
    },
    {
      "epoch": 5.402317219336796,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 13522
    },
    {
      "epoch": 5.402716739912106,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 13523
    },
    {
      "epoch": 5.4031162604874154,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 13524
    },
    {
      "epoch": 5.403515781062724,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 13525
    },
    {
      "epoch": 5.403915301638034,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 13526
    },
    {
      "epoch": 5.404314822213344,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 13527
    },
    {
      "epoch": 5.404714342788654,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 13528
    },
    {
      "epoch": 5.4051138633639635,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 13529
    },
    {
      "epoch": 5.405513383939273,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 13530
    },
    {
      "epoch": 5.405912904514582,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 13531
    },
    {
      "epoch": 5.406312425089892,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 13532
    },
    {
      "epoch": 5.406711945665202,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 13533
    },
    {
      "epoch": 5.4071114662405115,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 13534
    },
    {
      "epoch": 5.407510986815821,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 13535
    },
    {
      "epoch": 5.407910507391131,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 13536
    },
    {
      "epoch": 5.40831002796644,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 13537
    },
    {
      "epoch": 5.40870954854175,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 13538
    },
    {
      "epoch": 5.40910906911706,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 13539
    },
    {
      "epoch": 5.409508589692369,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 13540
    },
    {
      "epoch": 5.409908110267679,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 13541
    },
    {
      "epoch": 5.410307630842988,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 13542
    },
    {
      "epoch": 5.410707151418298,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 13543
    },
    {
      "epoch": 5.411106671993608,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 13544
    },
    {
      "epoch": 5.411506192568917,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 13545
    },
    {
      "epoch": 5.411905713144227,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 13546
    },
    {
      "epoch": 5.412305233719536,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 13547
    },
    {
      "epoch": 5.412704754294846,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 13548
    },
    {
      "epoch": 5.413104274870156,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 13549
    },
    {
      "epoch": 5.4135037954454654,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 13550
    },
    {
      "epoch": 5.413903316020775,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 13551
    },
    {
      "epoch": 5.414302836596085,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 13552
    },
    {
      "epoch": 5.414702357171394,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 13553
    },
    {
      "epoch": 5.415101877746704,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 13554
    },
    {
      "epoch": 5.4155013983220135,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 13555
    },
    {
      "epoch": 5.415900918897323,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 13556
    },
    {
      "epoch": 5.416300439472633,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 13557
    },
    {
      "epoch": 5.416699960047943,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 13558
    },
    {
      "epoch": 5.417099480623252,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 13559
    },
    {
      "epoch": 5.4174990011985615,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 13560
    },
    {
      "epoch": 5.417898521773871,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 13561
    },
    {
      "epoch": 5.418298042349181,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 13562
    },
    {
      "epoch": 5.418697562924491,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 13563
    },
    {
      "epoch": 5.4190970834998,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 13564
    },
    {
      "epoch": 5.41949660407511,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 13565
    },
    {
      "epoch": 5.419896124650419,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 13566
    },
    {
      "epoch": 5.420295645225729,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 13567
    },
    {
      "epoch": 5.420695165801039,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 13568
    },
    {
      "epoch": 5.421094686376349,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 13569
    },
    {
      "epoch": 5.421494206951658,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 13570
    },
    {
      "epoch": 5.421893727526967,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 13571
    },
    {
      "epoch": 5.422293248102277,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 13572
    },
    {
      "epoch": 5.422692768677587,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 13573
    },
    {
      "epoch": 5.423092289252897,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 13574
    },
    {
      "epoch": 5.4234918098282066,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 13575
    },
    {
      "epoch": 5.4238913304035155,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 13576
    },
    {
      "epoch": 5.424290850978825,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 13577
    },
    {
      "epoch": 5.424690371554135,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 13578
    },
    {
      "epoch": 5.425089892129445,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 13579
    },
    {
      "epoch": 5.425489412704755,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 13580
    },
    {
      "epoch": 5.4258889332800635,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 13581
    },
    {
      "epoch": 5.426288453855373,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 13582
    },
    {
      "epoch": 5.426687974430683,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 13583
    },
    {
      "epoch": 5.427087495005993,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 13584
    },
    {
      "epoch": 5.427487015581303,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 13585
    },
    {
      "epoch": 5.427886536156612,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 13586
    },
    {
      "epoch": 5.428286056731921,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 13587
    },
    {
      "epoch": 5.428685577307231,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 13588
    },
    {
      "epoch": 5.429085097882541,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 13589
    },
    {
      "epoch": 5.429484618457851,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 13590
    },
    {
      "epoch": 5.4298841390331605,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 13591
    },
    {
      "epoch": 5.43028365960847,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 13592
    },
    {
      "epoch": 5.430683180183779,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 13593
    },
    {
      "epoch": 5.431082700759089,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 13594
    },
    {
      "epoch": 5.431482221334399,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 13595
    },
    {
      "epoch": 5.4318817419097085,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 13596
    },
    {
      "epoch": 5.432281262485018,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 13597
    },
    {
      "epoch": 5.432680783060327,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 13598
    },
    {
      "epoch": 5.433080303635637,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 13599
    },
    {
      "epoch": 5.433479824210947,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 13600
    },
    {
      "epoch": 5.433879344786257,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 13601
    },
    {
      "epoch": 5.434278865361566,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 13602
    },
    {
      "epoch": 5.434678385936876,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 13603
    },
    {
      "epoch": 5.435077906512185,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 13604
    },
    {
      "epoch": 5.435477427087495,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 13605
    },
    {
      "epoch": 5.435876947662805,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 13606
    },
    {
      "epoch": 5.436276468238114,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 13607
    },
    {
      "epoch": 5.436675988813424,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 13608
    },
    {
      "epoch": 5.437075509388734,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 13609
    },
    {
      "epoch": 5.437475029964043,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 13610
    },
    {
      "epoch": 5.437874550539353,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 13611
    },
    {
      "epoch": 5.438274071114662,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 13612
    },
    {
      "epoch": 5.438673591689972,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 13613
    },
    {
      "epoch": 5.439073112265282,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 13614
    },
    {
      "epoch": 5.439472632840591,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 13615
    },
    {
      "epoch": 5.439872153415901,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 13616
    },
    {
      "epoch": 5.4402716739912105,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 13617
    },
    {
      "epoch": 5.44067119456652,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 13618
    },
    {
      "epoch": 5.44107071514183,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 13619
    },
    {
      "epoch": 5.44147023571714,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 13620
    },
    {
      "epoch": 5.441869756292449,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 13621
    },
    {
      "epoch": 5.4422692768677585,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 13622
    },
    {
      "epoch": 5.442668797443068,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 13623
    },
    {
      "epoch": 5.443068318018378,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 13624
    },
    {
      "epoch": 5.443467838593688,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 13625
    },
    {
      "epoch": 5.443867359168998,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 13626
    },
    {
      "epoch": 5.444266879744307,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 13627
    },
    {
      "epoch": 5.444666400319616,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 13628
    },
    {
      "epoch": 5.445065920894926,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 13629
    },
    {
      "epoch": 5.445465441470236,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 13630
    },
    {
      "epoch": 5.445864962045546,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 13631
    },
    {
      "epoch": 5.446264482620855,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 13632
    },
    {
      "epoch": 5.446664003196164,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 13633
    },
    {
      "epoch": 5.447063523771474,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 13634
    },
    {
      "epoch": 5.447463044346784,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 13635
    },
    {
      "epoch": 5.447862564922094,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 13636
    },
    {
      "epoch": 5.448262085497403,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 13637
    },
    {
      "epoch": 5.448661606072712,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 13638
    },
    {
      "epoch": 5.449061126648022,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 13639
    },
    {
      "epoch": 5.449460647223332,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 13640
    },
    {
      "epoch": 5.449860167798642,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 13641
    },
    {
      "epoch": 5.450259688373952,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 13642
    },
    {
      "epoch": 5.4506592089492605,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 13643
    },
    {
      "epoch": 5.45105872952457,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 13644
    },
    {
      "epoch": 5.45145825009988,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 13645
    },
    {
      "epoch": 5.45185777067519,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 13646
    },
    {
      "epoch": 5.4522572912505,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 13647
    },
    {
      "epoch": 5.452656811825809,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 13648
    },
    {
      "epoch": 5.453056332401118,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 13649
    },
    {
      "epoch": 5.453455852976428,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 13650
    },
    {
      "epoch": 5.453855373551738,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 13651
    },
    {
      "epoch": 5.454254894127048,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 13652
    },
    {
      "epoch": 5.4546544147023575,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 13653
    },
    {
      "epoch": 5.455053935277666,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 13654
    },
    {
      "epoch": 5.455453455852976,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 13655
    },
    {
      "epoch": 5.455852976428286,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 13656
    },
    {
      "epoch": 5.456252497003596,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 13657
    },
    {
      "epoch": 5.4566520175789055,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 13658
    },
    {
      "epoch": 5.457051538154215,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 13659
    },
    {
      "epoch": 5.457451058729524,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 13660
    },
    {
      "epoch": 5.457850579304834,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 13661
    },
    {
      "epoch": 5.458250099880144,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 13662
    },
    {
      "epoch": 5.4586496204554535,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 13663
    },
    {
      "epoch": 5.459049141030763,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 13664
    },
    {
      "epoch": 5.459448661606073,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 13665
    },
    {
      "epoch": 5.459848182181382,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 13666
    },
    {
      "epoch": 5.460247702756692,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 13667
    },
    {
      "epoch": 5.460647223332002,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 13668
    },
    {
      "epoch": 5.461046743907311,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 13669
    },
    {
      "epoch": 5.461446264482621,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 13670
    },
    {
      "epoch": 5.46184578505793,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 13671
    },
    {
      "epoch": 5.46224530563324,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 13672
    },
    {
      "epoch": 5.46264482620855,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 13673
    },
    {
      "epoch": 5.463044346783859,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 13674
    },
    {
      "epoch": 5.463443867359169,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 13675
    },
    {
      "epoch": 5.463843387934479,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 13676
    },
    {
      "epoch": 5.464242908509788,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 13677
    },
    {
      "epoch": 5.464642429085098,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 13678
    },
    {
      "epoch": 5.4650419496604075,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 13679
    },
    {
      "epoch": 5.465441470235717,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 13680
    },
    {
      "epoch": 5.465840990811027,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 13681
    },
    {
      "epoch": 5.466240511386337,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 13682
    },
    {
      "epoch": 5.466640031961646,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 13683
    },
    {
      "epoch": 5.4670395525369555,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 13684
    },
    {
      "epoch": 5.467439073112265,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 13685
    },
    {
      "epoch": 5.467838593687575,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 13686
    },
    {
      "epoch": 5.468238114262885,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 13687
    },
    {
      "epoch": 5.468637634838194,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 13688
    },
    {
      "epoch": 5.4690371554135035,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 13689
    },
    {
      "epoch": 5.469436675988813,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 13690
    },
    {
      "epoch": 5.469836196564123,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 13691
    },
    {
      "epoch": 5.470235717139433,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 13692
    },
    {
      "epoch": 5.470635237714743,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 13693
    },
    {
      "epoch": 5.471034758290052,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 13694
    },
    {
      "epoch": 5.471434278865361,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 13695
    },
    {
      "epoch": 5.471833799440671,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 13696
    },
    {
      "epoch": 5.472233320015981,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 13697
    },
    {
      "epoch": 5.472632840591291,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 13698
    },
    {
      "epoch": 5.4730323611666005,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 13699
    },
    {
      "epoch": 5.473431881741909,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 13700
    },
    {
      "epoch": 5.473831402317219,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 13701
    },
    {
      "epoch": 5.474230922892529,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 13702
    },
    {
      "epoch": 5.474630443467839,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 13703
    },
    {
      "epoch": 5.475029964043149,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 13704
    },
    {
      "epoch": 5.4754294846184575,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 13705
    },
    {
      "epoch": 5.475829005193767,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 13706
    },
    {
      "epoch": 5.476228525769077,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 13707
    },
    {
      "epoch": 5.476628046344387,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 13708
    },
    {
      "epoch": 5.477027566919697,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 13709
    },
    {
      "epoch": 5.477427087495006,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 13710
    },
    {
      "epoch": 5.477826608070315,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 13711
    },
    {
      "epoch": 5.478226128645625,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 13712
    },
    {
      "epoch": 5.478625649220935,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 13713
    },
    {
      "epoch": 5.479025169796245,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 13714
    },
    {
      "epoch": 5.479424690371554,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 13715
    },
    {
      "epoch": 5.479824210946864,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 13716
    },
    {
      "epoch": 5.480223731522173,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 13717
    },
    {
      "epoch": 5.480623252097483,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 13718
    },
    {
      "epoch": 5.481022772672793,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 13719
    },
    {
      "epoch": 5.4814222932481025,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 13720
    },
    {
      "epoch": 5.481821813823412,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 13721
    },
    {
      "epoch": 5.482221334398721,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 13722
    },
    {
      "epoch": 5.482620854974031,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 13723
    },
    {
      "epoch": 5.483020375549341,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 13724
    },
    {
      "epoch": 5.4834198961246505,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 13725
    },
    {
      "epoch": 5.48381941669996,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0598,
      "step": 13726
    },
    {
      "epoch": 5.484218937275269,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 13727
    },
    {
      "epoch": 5.484618457850579,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 13728
    },
    {
      "epoch": 5.485017978425889,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 13729
    },
    {
      "epoch": 5.485417499001199,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 13730
    },
    {
      "epoch": 5.485817019576508,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 13731
    },
    {
      "epoch": 5.486216540151818,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 13732
    },
    {
      "epoch": 5.486616060727127,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 13733
    },
    {
      "epoch": 5.487015581302437,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 13734
    },
    {
      "epoch": 5.487415101877747,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 13735
    },
    {
      "epoch": 5.487814622453056,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 13736
    },
    {
      "epoch": 5.488214143028366,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 13737
    },
    {
      "epoch": 5.488613663603676,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 13738
    },
    {
      "epoch": 5.489013184178985,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 13739
    },
    {
      "epoch": 5.489412704754295,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 13740
    },
    {
      "epoch": 5.4898122253296044,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 13741
    },
    {
      "epoch": 5.490211745904914,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 13742
    },
    {
      "epoch": 5.490611266480224,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 13743
    },
    {
      "epoch": 5.491010787055533,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 13744
    },
    {
      "epoch": 5.491410307630843,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 13745
    },
    {
      "epoch": 5.4918098282061525,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 13746
    },
    {
      "epoch": 5.492209348781462,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 13747
    },
    {
      "epoch": 5.492608869356772,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 13748
    },
    {
      "epoch": 5.493008389932082,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 13749
    },
    {
      "epoch": 5.493407910507391,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 13750
    },
    {
      "epoch": 5.4938074310827005,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 13751
    },
    {
      "epoch": 5.49420695165801,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 13752
    },
    {
      "epoch": 5.49460647223332,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 13753
    },
    {
      "epoch": 5.49500599280863,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 13754
    },
    {
      "epoch": 5.49540551338394,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 13755
    },
    {
      "epoch": 5.495805033959249,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 13756
    },
    {
      "epoch": 5.496204554534558,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 13757
    },
    {
      "epoch": 5.496604075109868,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 13758
    },
    {
      "epoch": 5.497003595685178,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 13759
    },
    {
      "epoch": 5.497403116260488,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 13760
    },
    {
      "epoch": 5.497802636835797,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0633,
      "step": 13761
    },
    {
      "epoch": 5.498202157411106,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 13762
    },
    {
      "epoch": 5.498601677986416,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 13763
    },
    {
      "epoch": 5.499001198561726,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 13764
    },
    {
      "epoch": 5.499400719137036,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 13765
    },
    {
      "epoch": 5.4998002397123456,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 13766
    },
    {
      "epoch": 5.5001997602876544,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 13767
    },
    {
      "epoch": 5.500599280862964,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 13768
    },
    {
      "epoch": 5.500998801438274,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 13769
    },
    {
      "epoch": 5.501398322013584,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 13770
    },
    {
      "epoch": 5.501797842588894,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 13771
    },
    {
      "epoch": 5.502197363164203,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 13772
    },
    {
      "epoch": 5.502596883739512,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 13773
    },
    {
      "epoch": 5.502996404314822,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 13774
    },
    {
      "epoch": 5.503395924890132,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0624,
      "step": 13775
    },
    {
      "epoch": 5.503795445465442,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 13776
    },
    {
      "epoch": 5.504194966040751,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 13777
    },
    {
      "epoch": 5.50459448661606,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 13778
    },
    {
      "epoch": 5.50499400719137,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 13779
    },
    {
      "epoch": 5.50539352776668,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 13780
    },
    {
      "epoch": 5.50579304834199,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 13781
    },
    {
      "epoch": 5.5061925689172995,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 13782
    },
    {
      "epoch": 5.506592089492609,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 13783
    },
    {
      "epoch": 5.506991610067918,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 13784
    },
    {
      "epoch": 5.507391130643228,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 13785
    },
    {
      "epoch": 5.507790651218538,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 13786
    },
    {
      "epoch": 5.5081901717938475,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 13787
    },
    {
      "epoch": 5.508589692369157,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 13788
    },
    {
      "epoch": 5.508989212944467,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 13789
    },
    {
      "epoch": 5.509388733519776,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 13790
    },
    {
      "epoch": 5.509788254095086,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 13791
    },
    {
      "epoch": 5.5101877746703956,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 13792
    },
    {
      "epoch": 5.510587295245705,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 13793
    },
    {
      "epoch": 5.510986815821015,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 13794
    },
    {
      "epoch": 5.511386336396324,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 13795
    },
    {
      "epoch": 5.511785856971634,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 13796
    },
    {
      "epoch": 5.512185377546944,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 13797
    },
    {
      "epoch": 5.512584898122253,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 13798
    },
    {
      "epoch": 5.512984418697563,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 13799
    },
    {
      "epoch": 5.513383939272872,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 13800
    },
    {
      "epoch": 5.513783459848182,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 13801
    },
    {
      "epoch": 5.514182980423492,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 13802
    },
    {
      "epoch": 5.514582500998801,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 13803
    },
    {
      "epoch": 5.514982021574111,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 13804
    },
    {
      "epoch": 5.515381542149421,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 13805
    },
    {
      "epoch": 5.515781062724731,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 13806
    },
    {
      "epoch": 5.51618058330004,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 13807
    },
    {
      "epoch": 5.5165801038753495,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 13808
    },
    {
      "epoch": 5.516979624450659,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 13809
    },
    {
      "epoch": 5.517379145025969,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 13810
    },
    {
      "epoch": 5.517778665601279,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 13811
    },
    {
      "epoch": 5.518178186176588,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 13812
    },
    {
      "epoch": 5.5185777067518975,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 13813
    },
    {
      "epoch": 5.518977227327207,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 13814
    },
    {
      "epoch": 5.519376747902517,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 13815
    },
    {
      "epoch": 5.519776268477827,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 13816
    },
    {
      "epoch": 5.520175789053136,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 13817
    },
    {
      "epoch": 5.520575309628446,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 13818
    },
    {
      "epoch": 5.520974830203755,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 13819
    },
    {
      "epoch": 5.521374350779065,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 13820
    },
    {
      "epoch": 5.521773871354375,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 13821
    },
    {
      "epoch": 5.522173391929685,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 13822
    },
    {
      "epoch": 5.5225729125049945,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0593,
      "step": 13823
    },
    {
      "epoch": 5.522972433080303,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 13824
    },
    {
      "epoch": 5.523371953655613,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 13825
    },
    {
      "epoch": 5.523771474230923,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 13826
    },
    {
      "epoch": 5.524170994806233,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 13827
    },
    {
      "epoch": 5.5245705153815425,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 13828
    },
    {
      "epoch": 5.524970035956851,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 13829
    },
    {
      "epoch": 5.525369556532161,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 13830
    },
    {
      "epoch": 5.525769077107471,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 13831
    },
    {
      "epoch": 5.526168597682781,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 13832
    },
    {
      "epoch": 5.526568118258091,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 13833
    },
    {
      "epoch": 5.5269676388333995,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 13834
    },
    {
      "epoch": 5.527367159408709,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 13835
    },
    {
      "epoch": 5.527766679984019,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 13836
    },
    {
      "epoch": 5.528166200559329,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 13837
    },
    {
      "epoch": 5.528565721134639,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 13838
    },
    {
      "epoch": 5.528965241709948,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 13839
    },
    {
      "epoch": 5.529364762285257,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 13840
    },
    {
      "epoch": 5.529764282860567,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 13841
    },
    {
      "epoch": 5.530163803435877,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 13842
    },
    {
      "epoch": 5.530563324011187,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 13843
    },
    {
      "epoch": 5.5309628445864965,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 13844
    },
    {
      "epoch": 5.531362365161806,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 13845
    },
    {
      "epoch": 5.531761885737115,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 13846
    },
    {
      "epoch": 5.532161406312425,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 13847
    },
    {
      "epoch": 5.532560926887735,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 13848
    },
    {
      "epoch": 5.5329604474630445,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 13849
    },
    {
      "epoch": 5.533359968038354,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 13850
    },
    {
      "epoch": 5.533759488613663,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 13851
    },
    {
      "epoch": 5.534159009188973,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 13852
    },
    {
      "epoch": 5.534558529764283,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 13853
    },
    {
      "epoch": 5.5349580503395925,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 13854
    },
    {
      "epoch": 5.535357570914902,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 13855
    },
    {
      "epoch": 5.535757091490212,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 13856
    },
    {
      "epoch": 5.536156612065521,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 13857
    },
    {
      "epoch": 5.536556132640831,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 13858
    },
    {
      "epoch": 5.536955653216141,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 13859
    },
    {
      "epoch": 5.53735517379145,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 13860
    },
    {
      "epoch": 5.53775469436676,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 13861
    },
    {
      "epoch": 5.53815421494207,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 13862
    },
    {
      "epoch": 5.538553735517379,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 13863
    },
    {
      "epoch": 5.538953256092689,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 13864
    },
    {
      "epoch": 5.539352776667998,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 13865
    },
    {
      "epoch": 5.539752297243308,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 13866
    },
    {
      "epoch": 5.540151817818618,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 13867
    },
    {
      "epoch": 5.540551338393927,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 13868
    },
    {
      "epoch": 5.540950858969237,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 13869
    },
    {
      "epoch": 5.5413503795445465,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 13870
    },
    {
      "epoch": 5.541749900119856,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 13871
    },
    {
      "epoch": 5.542149420695166,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 13872
    },
    {
      "epoch": 5.542548941270476,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 13873
    },
    {
      "epoch": 5.542948461845785,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 13874
    },
    {
      "epoch": 5.5433479824210945,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 13875
    },
    {
      "epoch": 5.543747502996404,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0587,
      "step": 13876
    },
    {
      "epoch": 5.544147023571714,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 13877
    },
    {
      "epoch": 5.544546544147024,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 13878
    },
    {
      "epoch": 5.544946064722334,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 13879
    },
    {
      "epoch": 5.5453455852976425,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 13880
    },
    {
      "epoch": 5.545745105872952,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 13881
    },
    {
      "epoch": 5.546144626448262,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 13882
    },
    {
      "epoch": 5.546544147023572,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 13883
    },
    {
      "epoch": 5.546943667598882,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 13884
    },
    {
      "epoch": 5.547343188174191,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 13885
    },
    {
      "epoch": 5.5477427087495,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 13886
    },
    {
      "epoch": 5.54814222932481,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 13887
    },
    {
      "epoch": 5.54854174990012,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 13888
    },
    {
      "epoch": 5.54894127047543,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 13889
    },
    {
      "epoch": 5.549340791050739,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 13890
    },
    {
      "epoch": 5.549740311626048,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 13891
    },
    {
      "epoch": 5.550139832201358,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 13892
    },
    {
      "epoch": 5.550539352776668,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 13893
    },
    {
      "epoch": 5.550938873351978,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 13894
    },
    {
      "epoch": 5.551338393927288,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.061,
      "step": 13895
    },
    {
      "epoch": 5.551737914502597,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 13896
    },
    {
      "epoch": 5.552137435077906,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 13897
    },
    {
      "epoch": 5.552536955653216,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 13898
    },
    {
      "epoch": 5.552936476228526,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 13899
    },
    {
      "epoch": 5.553335996803836,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 13900
    },
    {
      "epoch": 5.553735517379145,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 13901
    },
    {
      "epoch": 5.554135037954454,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 13902
    },
    {
      "epoch": 5.554534558529764,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 13903
    },
    {
      "epoch": 5.554934079105074,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 13904
    },
    {
      "epoch": 5.555333599680384,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 13905
    },
    {
      "epoch": 5.555733120255693,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 13906
    },
    {
      "epoch": 5.556132640831002,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 13907
    },
    {
      "epoch": 5.556532161406312,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 13908
    },
    {
      "epoch": 5.556931681981622,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 13909
    },
    {
      "epoch": 5.557331202556932,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 13910
    },
    {
      "epoch": 5.5577307231322415,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 13911
    },
    {
      "epoch": 5.558130243707551,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 13912
    },
    {
      "epoch": 5.558529764282861,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 13913
    },
    {
      "epoch": 5.55892928485817,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 13914
    },
    {
      "epoch": 5.55932880543348,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 13915
    },
    {
      "epoch": 5.5597283260087895,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 13916
    },
    {
      "epoch": 5.560127846584099,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 13917
    },
    {
      "epoch": 5.560527367159409,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 13918
    },
    {
      "epoch": 5.560926887734718,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 13919
    },
    {
      "epoch": 5.561326408310028,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 13920
    },
    {
      "epoch": 5.561725928885338,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 13921
    },
    {
      "epoch": 5.562125449460647,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 13922
    },
    {
      "epoch": 5.562524970035957,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 13923
    },
    {
      "epoch": 5.562924490611266,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 13924
    },
    {
      "epoch": 5.563324011186576,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 13925
    },
    {
      "epoch": 5.563723531761886,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 13926
    },
    {
      "epoch": 5.564123052337195,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0574,
      "step": 13927
    },
    {
      "epoch": 5.564522572912505,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 13928
    },
    {
      "epoch": 5.564922093487815,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 13929
    },
    {
      "epoch": 5.565321614063124,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 13930
    },
    {
      "epoch": 5.565721134638434,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 13931
    },
    {
      "epoch": 5.566120655213743,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 13932
    },
    {
      "epoch": 5.566520175789053,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 13933
    },
    {
      "epoch": 5.566919696364363,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 13934
    },
    {
      "epoch": 5.567319216939673,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 13935
    },
    {
      "epoch": 5.567718737514982,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 13936
    },
    {
      "epoch": 5.5681182580902915,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 13937
    },
    {
      "epoch": 5.568517778665601,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 13938
    },
    {
      "epoch": 5.568917299240911,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 13939
    },
    {
      "epoch": 5.569316819816221,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 13940
    },
    {
      "epoch": 5.56971634039153,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 13941
    },
    {
      "epoch": 5.5701158609668395,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 13942
    },
    {
      "epoch": 5.570515381542149,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 13943
    },
    {
      "epoch": 5.570914902117459,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 13944
    },
    {
      "epoch": 5.571314422692769,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 13945
    },
    {
      "epoch": 5.571713943268079,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 13946
    },
    {
      "epoch": 5.572113463843388,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 13947
    },
    {
      "epoch": 5.572512984418697,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 13948
    },
    {
      "epoch": 5.572912504994007,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 13949
    },
    {
      "epoch": 5.573312025569317,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 13950
    },
    {
      "epoch": 5.573711546144627,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 13951
    },
    {
      "epoch": 5.5741110667199365,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 13952
    },
    {
      "epoch": 5.574510587295245,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 13953
    },
    {
      "epoch": 5.574910107870555,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 13954
    },
    {
      "epoch": 5.575309628445865,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 13955
    },
    {
      "epoch": 5.575709149021175,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 13956
    },
    {
      "epoch": 5.5761086695964845,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 13957
    },
    {
      "epoch": 5.5765081901717934,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 13958
    },
    {
      "epoch": 5.576907710747103,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 13959
    },
    {
      "epoch": 5.577307231322413,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 13960
    },
    {
      "epoch": 5.577706751897723,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 13961
    },
    {
      "epoch": 5.578106272473033,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 13962
    },
    {
      "epoch": 5.578505793048342,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 13963
    },
    {
      "epoch": 5.578905313623651,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 13964
    },
    {
      "epoch": 5.579304834198961,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 13965
    },
    {
      "epoch": 5.579704354774271,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 13966
    },
    {
      "epoch": 5.580103875349581,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 13967
    },
    {
      "epoch": 5.58050339592489,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 13968
    },
    {
      "epoch": 5.5809029165002,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 13969
    },
    {
      "epoch": 5.581302437075509,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 13970
    },
    {
      "epoch": 5.581701957650819,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 13971
    },
    {
      "epoch": 5.582101478226129,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 13972
    },
    {
      "epoch": 5.5825009988014385,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 13973
    },
    {
      "epoch": 5.582900519376748,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 13974
    },
    {
      "epoch": 5.583300039952057,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 13975
    },
    {
      "epoch": 5.583699560527367,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 13976
    },
    {
      "epoch": 5.584099081102677,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 13977
    },
    {
      "epoch": 5.5844986016779865,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 13978
    },
    {
      "epoch": 5.584898122253296,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 13979
    },
    {
      "epoch": 5.585297642828605,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 13980
    },
    {
      "epoch": 5.585697163403915,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 13981
    },
    {
      "epoch": 5.586096683979225,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 13982
    },
    {
      "epoch": 5.5864962045545346,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 13983
    },
    {
      "epoch": 5.586895725129844,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 13984
    },
    {
      "epoch": 5.587295245705154,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 13985
    },
    {
      "epoch": 5.587694766280464,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 13986
    },
    {
      "epoch": 5.588094286855773,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 13987
    },
    {
      "epoch": 5.588493807431083,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 13988
    },
    {
      "epoch": 5.588893328006392,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 13989
    },
    {
      "epoch": 5.589292848581702,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 13990
    },
    {
      "epoch": 5.589692369157012,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 13991
    },
    {
      "epoch": 5.590091889732321,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 13992
    },
    {
      "epoch": 5.590491410307631,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 13993
    },
    {
      "epoch": 5.59089093088294,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 13994
    },
    {
      "epoch": 5.59129045145825,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 13995
    },
    {
      "epoch": 5.59168997203356,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 13996
    },
    {
      "epoch": 5.592089492608869,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 13997
    },
    {
      "epoch": 5.592489013184179,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 13998
    },
    {
      "epoch": 5.5928885337594885,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 13999
    },
    {
      "epoch": 5.593288054334798,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 14000
    },
    {
      "epoch": 5.593687574910108,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 14001
    },
    {
      "epoch": 5.594087095485418,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 14002
    },
    {
      "epoch": 5.594486616060728,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 14003
    },
    {
      "epoch": 5.5948861366360365,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 14004
    },
    {
      "epoch": 5.595285657211346,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 14005
    },
    {
      "epoch": 5.595685177786656,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 14006
    },
    {
      "epoch": 5.596084698361966,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 14007
    },
    {
      "epoch": 5.596484218937276,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 14008
    },
    {
      "epoch": 5.5968837395125846,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 14009
    },
    {
      "epoch": 5.597283260087894,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 14010
    },
    {
      "epoch": 5.597682780663204,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 14011
    },
    {
      "epoch": 5.598082301238514,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 14012
    },
    {
      "epoch": 5.598481821813824,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 14013
    },
    {
      "epoch": 5.598881342389133,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 14014
    },
    {
      "epoch": 5.599280862964442,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 14015
    },
    {
      "epoch": 5.599680383539752,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 14016
    },
    {
      "epoch": 5.600079904115062,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 14017
    },
    {
      "epoch": 5.600479424690372,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 14018
    },
    {
      "epoch": 5.6008789452656815,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 14019
    },
    {
      "epoch": 5.60127846584099,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 14020
    },
    {
      "epoch": 5.6016779864163,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 14021
    },
    {
      "epoch": 5.60207750699161,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 14022
    },
    {
      "epoch": 5.60247702756692,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 14023
    },
    {
      "epoch": 5.60287654814223,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 14024
    },
    {
      "epoch": 5.603276068717539,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 14025
    },
    {
      "epoch": 5.603675589292848,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 14026
    },
    {
      "epoch": 5.604075109868158,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 14027
    },
    {
      "epoch": 5.604474630443468,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 14028
    },
    {
      "epoch": 5.604874151018778,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 14029
    },
    {
      "epoch": 5.605273671594087,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 14030
    },
    {
      "epoch": 5.605673192169396,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 14031
    },
    {
      "epoch": 5.606072712744706,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 14032
    },
    {
      "epoch": 5.606472233320016,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 14033
    },
    {
      "epoch": 5.606871753895326,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 14034
    },
    {
      "epoch": 5.6072712744706354,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 14035
    },
    {
      "epoch": 5.607670795045945,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 14036
    },
    {
      "epoch": 5.608070315621254,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 14037
    },
    {
      "epoch": 5.608469836196564,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 14038
    },
    {
      "epoch": 5.608869356771874,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 14039
    },
    {
      "epoch": 5.6092688773471835,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 14040
    },
    {
      "epoch": 5.609668397922493,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 14041
    },
    {
      "epoch": 5.610067918497803,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 14042
    },
    {
      "epoch": 5.610467439073112,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 14043
    },
    {
      "epoch": 5.610866959648422,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 14044
    },
    {
      "epoch": 5.6112664802237315,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 14045
    },
    {
      "epoch": 5.611666000799041,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 14046
    },
    {
      "epoch": 5.612065521374351,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 14047
    },
    {
      "epoch": 5.61246504194966,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 14048
    },
    {
      "epoch": 5.61286456252497,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 14049
    },
    {
      "epoch": 5.61326408310028,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 14050
    },
    {
      "epoch": 5.613663603675589,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 14051
    },
    {
      "epoch": 5.614063124250899,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 14052
    },
    {
      "epoch": 5.614462644826209,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 14053
    },
    {
      "epoch": 5.614862165401518,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 14054
    },
    {
      "epoch": 5.615261685976828,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 14055
    },
    {
      "epoch": 5.615661206552137,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 14056
    },
    {
      "epoch": 5.616060727127447,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 14057
    },
    {
      "epoch": 5.616460247702757,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 14058
    },
    {
      "epoch": 5.616859768278067,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 14059
    },
    {
      "epoch": 5.617259288853376,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 14060
    },
    {
      "epoch": 5.6176588094286855,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 14061
    },
    {
      "epoch": 5.618058330003995,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 14062
    },
    {
      "epoch": 5.618457850579305,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 14063
    },
    {
      "epoch": 5.618857371154615,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 14064
    },
    {
      "epoch": 5.619256891729924,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 14065
    },
    {
      "epoch": 5.6196564123052335,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 14066
    },
    {
      "epoch": 5.620055932880543,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 14067
    },
    {
      "epoch": 5.620455453455853,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 14068
    },
    {
      "epoch": 5.620854974031163,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 14069
    },
    {
      "epoch": 5.621254494606472,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 14070
    },
    {
      "epoch": 5.6216540151817815,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 14071
    },
    {
      "epoch": 5.622053535757091,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 14072
    },
    {
      "epoch": 5.622453056332401,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 14073
    },
    {
      "epoch": 5.622852576907711,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 14074
    },
    {
      "epoch": 5.623252097483021,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 14075
    },
    {
      "epoch": 5.6236516180583305,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 14076
    },
    {
      "epoch": 5.624051138633639,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 14077
    },
    {
      "epoch": 5.624450659208949,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 14078
    },
    {
      "epoch": 5.624850179784259,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 14079
    },
    {
      "epoch": 5.625249700359569,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 14080
    },
    {
      "epoch": 5.6256492209348785,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 14081
    },
    {
      "epoch": 5.626048741510187,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 14082
    },
    {
      "epoch": 5.626448262085497,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 14083
    },
    {
      "epoch": 5.626847782660807,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 14084
    },
    {
      "epoch": 5.627247303236117,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 14085
    },
    {
      "epoch": 5.627646823811427,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 14086
    },
    {
      "epoch": 5.6280463443867355,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 14087
    },
    {
      "epoch": 5.628445864962045,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 14088
    },
    {
      "epoch": 5.628845385537355,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 14089
    },
    {
      "epoch": 5.629244906112665,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 14090
    },
    {
      "epoch": 5.629644426687975,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 14091
    },
    {
      "epoch": 5.630043947263284,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 14092
    },
    {
      "epoch": 5.630443467838594,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 14093
    },
    {
      "epoch": 5.630842988413903,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 14094
    },
    {
      "epoch": 5.631242508989213,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 14095
    },
    {
      "epoch": 5.631642029564523,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 14096
    },
    {
      "epoch": 5.632041550139832,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 14097
    },
    {
      "epoch": 5.632441070715142,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 14098
    },
    {
      "epoch": 5.632840591290451,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 14099
    },
    {
      "epoch": 5.633240111865761,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 14100
    },
    {
      "epoch": 5.633639632441071,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 14101
    },
    {
      "epoch": 5.6340391530163805,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 14102
    },
    {
      "epoch": 5.63443867359169,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 14103
    },
    {
      "epoch": 5.634838194166999,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 14104
    },
    {
      "epoch": 5.635237714742309,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 14105
    },
    {
      "epoch": 5.635637235317619,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 14106
    },
    {
      "epoch": 5.6360367558929285,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 14107
    },
    {
      "epoch": 5.636436276468238,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 14108
    },
    {
      "epoch": 5.636835797043548,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 14109
    },
    {
      "epoch": 5.637235317618857,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 14110
    },
    {
      "epoch": 5.637634838194167,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 14111
    },
    {
      "epoch": 5.638034358769477,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 14112
    },
    {
      "epoch": 5.638433879344786,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0604,
      "step": 14113
    },
    {
      "epoch": 5.638833399920096,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 14114
    },
    {
      "epoch": 5.639232920495406,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 14115
    },
    {
      "epoch": 5.639632441070715,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 14116
    },
    {
      "epoch": 5.640031961646025,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 14117
    },
    {
      "epoch": 5.640431482221334,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 14118
    },
    {
      "epoch": 5.640831002796644,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 14119
    },
    {
      "epoch": 5.641230523371954,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 14120
    },
    {
      "epoch": 5.641630043947263,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 14121
    },
    {
      "epoch": 5.642029564522573,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 14122
    },
    {
      "epoch": 5.642429085097882,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 14123
    },
    {
      "epoch": 5.642828605673192,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 14124
    },
    {
      "epoch": 5.643228126248502,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 14125
    },
    {
      "epoch": 5.643627646823812,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 14126
    },
    {
      "epoch": 5.644027167399121,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 14127
    },
    {
      "epoch": 5.6444266879744305,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 14128
    },
    {
      "epoch": 5.64482620854974,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 14129
    },
    {
      "epoch": 5.64522572912505,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 14130
    },
    {
      "epoch": 5.64562524970036,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 14131
    },
    {
      "epoch": 5.64602477027567,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 14132
    },
    {
      "epoch": 5.6464242908509785,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 14133
    },
    {
      "epoch": 5.646823811426288,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 14134
    },
    {
      "epoch": 5.647223332001598,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 14135
    },
    {
      "epoch": 5.647622852576908,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 14136
    },
    {
      "epoch": 5.648022373152218,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 14137
    },
    {
      "epoch": 5.648421893727527,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 14138
    },
    {
      "epoch": 5.648821414302836,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 14139
    },
    {
      "epoch": 5.649220934878146,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 14140
    },
    {
      "epoch": 5.649620455453456,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 14141
    },
    {
      "epoch": 5.650019976028766,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 14142
    },
    {
      "epoch": 5.6504194966040755,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 14143
    },
    {
      "epoch": 5.650819017179384,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 14144
    },
    {
      "epoch": 5.651218537754694,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 14145
    },
    {
      "epoch": 5.651618058330004,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 14146
    },
    {
      "epoch": 5.652017578905314,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 14147
    },
    {
      "epoch": 5.6524170994806235,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 14148
    },
    {
      "epoch": 5.652816620055933,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 14149
    },
    {
      "epoch": 5.653216140631242,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 14150
    },
    {
      "epoch": 5.653615661206552,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 14151
    },
    {
      "epoch": 5.654015181781862,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 14152
    },
    {
      "epoch": 5.654414702357172,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 14153
    },
    {
      "epoch": 5.654814222932481,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 14154
    },
    {
      "epoch": 5.65521374350779,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 14155
    },
    {
      "epoch": 5.6556132640831,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 14156
    },
    {
      "epoch": 5.65601278465841,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 14157
    },
    {
      "epoch": 5.65641230523372,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 14158
    },
    {
      "epoch": 5.656811825809029,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 14159
    },
    {
      "epoch": 5.657211346384338,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 14160
    },
    {
      "epoch": 5.657610866959648,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 14161
    },
    {
      "epoch": 5.658010387534958,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 14162
    },
    {
      "epoch": 5.658409908110268,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 14163
    },
    {
      "epoch": 5.6588094286855775,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 14164
    },
    {
      "epoch": 5.659208949260887,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 14165
    },
    {
      "epoch": 5.659608469836197,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 14166
    },
    {
      "epoch": 5.660007990411506,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 14167
    },
    {
      "epoch": 5.660407510986816,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 14168
    },
    {
      "epoch": 5.6608070315621255,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 14169
    },
    {
      "epoch": 5.661206552137435,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 14170
    },
    {
      "epoch": 5.661606072712745,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 14171
    },
    {
      "epoch": 5.662005593288054,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 14172
    },
    {
      "epoch": 5.662405113863364,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 14173
    },
    {
      "epoch": 5.6628046344386735,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 14174
    },
    {
      "epoch": 5.663204155013983,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 14175
    },
    {
      "epoch": 5.663603675589293,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 14176
    },
    {
      "epoch": 5.664003196164602,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 14177
    },
    {
      "epoch": 5.664402716739912,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 14178
    },
    {
      "epoch": 5.664802237315222,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 14179
    },
    {
      "epoch": 5.665201757890531,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 14180
    },
    {
      "epoch": 5.665601278465841,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 14181
    },
    {
      "epoch": 5.666000799041151,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 14182
    },
    {
      "epoch": 5.666400319616461,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 14183
    },
    {
      "epoch": 5.66679984019177,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 14184
    },
    {
      "epoch": 5.667199360767079,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 14185
    },
    {
      "epoch": 5.667598881342389,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 14186
    },
    {
      "epoch": 5.667998401917699,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 14187
    },
    {
      "epoch": 5.668397922493009,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 14188
    },
    {
      "epoch": 5.668797443068318,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 14189
    },
    {
      "epoch": 5.6691969636436275,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 14190
    },
    {
      "epoch": 5.669596484218937,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 14191
    },
    {
      "epoch": 5.669996004794247,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 14192
    },
    {
      "epoch": 5.670395525369557,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 14193
    },
    {
      "epoch": 5.670795045944866,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 14194
    },
    {
      "epoch": 5.6711945665201755,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 14195
    },
    {
      "epoch": 5.671594087095485,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 14196
    },
    {
      "epoch": 5.671993607670795,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 14197
    },
    {
      "epoch": 5.672393128246105,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 14198
    },
    {
      "epoch": 5.672792648821415,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 14199
    },
    {
      "epoch": 5.6731921693967235,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 14200
    },
    {
      "epoch": 5.673591689972033,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 14201
    },
    {
      "epoch": 5.673991210547343,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 14202
    },
    {
      "epoch": 5.674390731122653,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 14203
    },
    {
      "epoch": 5.674790251697963,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 14204
    },
    {
      "epoch": 5.6751897722732725,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 14205
    },
    {
      "epoch": 5.675589292848581,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 14206
    },
    {
      "epoch": 5.675988813423891,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 14207
    },
    {
      "epoch": 5.676388333999201,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 14208
    },
    {
      "epoch": 5.676787854574511,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 14209
    },
    {
      "epoch": 5.6771873751498205,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 14210
    },
    {
      "epoch": 5.677586895725129,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 14211
    },
    {
      "epoch": 5.677986416300439,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 14212
    },
    {
      "epoch": 5.678385936875749,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 14213
    },
    {
      "epoch": 5.678785457451059,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 14214
    },
    {
      "epoch": 5.679184978026369,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 14215
    },
    {
      "epoch": 5.679584498601678,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 14216
    },
    {
      "epoch": 5.679984019176987,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 14217
    },
    {
      "epoch": 5.680383539752297,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 14218
    },
    {
      "epoch": 5.680783060327607,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 14219
    },
    {
      "epoch": 5.681182580902917,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 14220
    },
    {
      "epoch": 5.681582101478226,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 14221
    },
    {
      "epoch": 5.681981622053536,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 14222
    },
    {
      "epoch": 5.682381142628845,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 14223
    },
    {
      "epoch": 5.682780663204155,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 14224
    },
    {
      "epoch": 5.683180183779465,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 14225
    },
    {
      "epoch": 5.683579704354774,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 14226
    },
    {
      "epoch": 5.683979224930084,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 14227
    },
    {
      "epoch": 5.684378745505393,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 14228
    },
    {
      "epoch": 5.684778266080703,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 14229
    },
    {
      "epoch": 5.685177786656013,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 14230
    },
    {
      "epoch": 5.6855773072313225,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 14231
    },
    {
      "epoch": 5.685976827806632,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 14232
    },
    {
      "epoch": 5.686376348381942,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 14233
    },
    {
      "epoch": 5.686775868957251,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 14234
    },
    {
      "epoch": 5.687175389532561,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 14235
    },
    {
      "epoch": 5.6875749101078705,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 14236
    },
    {
      "epoch": 5.68797443068318,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 14237
    },
    {
      "epoch": 5.68837395125849,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 14238
    },
    {
      "epoch": 5.6887734718338,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 14239
    },
    {
      "epoch": 5.689172992409109,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 14240
    },
    {
      "epoch": 5.689572512984419,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 14241
    },
    {
      "epoch": 5.689972033559728,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 14242
    },
    {
      "epoch": 5.690371554135038,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 14243
    },
    {
      "epoch": 5.690771074710348,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 14244
    },
    {
      "epoch": 5.691170595285657,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 14245
    },
    {
      "epoch": 5.691570115860967,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 14246
    },
    {
      "epoch": 5.691969636436276,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 14247
    },
    {
      "epoch": 5.692369157011586,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 14248
    },
    {
      "epoch": 5.692768677586896,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 14249
    },
    {
      "epoch": 5.693168198162205,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 14250
    },
    {
      "epoch": 5.693567718737515,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 14251
    },
    {
      "epoch": 5.6939672393128244,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 14252
    },
    {
      "epoch": 5.694366759888134,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 14253
    },
    {
      "epoch": 5.694766280463444,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 14254
    },
    {
      "epoch": 5.695165801038754,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 14255
    },
    {
      "epoch": 5.695565321614064,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 14256
    },
    {
      "epoch": 5.6959648421893725,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 14257
    },
    {
      "epoch": 5.696364362764682,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 14258
    },
    {
      "epoch": 5.696763883339992,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 14259
    },
    {
      "epoch": 5.697163403915302,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 14260
    },
    {
      "epoch": 5.697562924490612,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 14261
    },
    {
      "epoch": 5.6979624450659205,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 14262
    },
    {
      "epoch": 5.69836196564123,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 14263
    },
    {
      "epoch": 5.69876148621654,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 14264
    },
    {
      "epoch": 5.69916100679185,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 14265
    },
    {
      "epoch": 5.69956052736716,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 14266
    },
    {
      "epoch": 5.699960047942469,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 14267
    },
    {
      "epoch": 5.700359568517778,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 14268
    },
    {
      "epoch": 5.700759089093088,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 14269
    },
    {
      "epoch": 5.701158609668398,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 14270
    },
    {
      "epoch": 5.701558130243708,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 14271
    },
    {
      "epoch": 5.7019576508190175,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 14272
    },
    {
      "epoch": 5.702357171394327,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 14273
    },
    {
      "epoch": 5.702756691969636,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 14274
    },
    {
      "epoch": 5.703156212544946,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 14275
    },
    {
      "epoch": 5.703555733120256,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 14276
    },
    {
      "epoch": 5.7039552536955656,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 14277
    },
    {
      "epoch": 5.704354774270875,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 14278
    },
    {
      "epoch": 5.704754294846184,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 14279
    },
    {
      "epoch": 5.705153815421494,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 14280
    },
    {
      "epoch": 5.705553335996804,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 14281
    },
    {
      "epoch": 5.705952856572114,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 14282
    },
    {
      "epoch": 5.706352377147423,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 14283
    },
    {
      "epoch": 5.706751897722732,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 14284
    },
    {
      "epoch": 5.707151418298042,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 14285
    },
    {
      "epoch": 5.707550938873352,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 14286
    },
    {
      "epoch": 5.707950459448662,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 14287
    },
    {
      "epoch": 5.708349980023971,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 14288
    },
    {
      "epoch": 5.708749500599281,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 14289
    },
    {
      "epoch": 5.70914902117459,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 14290
    },
    {
      "epoch": 5.7095485417499,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 14291
    },
    {
      "epoch": 5.70994806232521,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 14292
    },
    {
      "epoch": 5.7103475829005195,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 14293
    },
    {
      "epoch": 5.710747103475829,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 14294
    },
    {
      "epoch": 5.711146624051139,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 14295
    },
    {
      "epoch": 5.711546144626448,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 14296
    },
    {
      "epoch": 5.711945665201758,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 14297
    },
    {
      "epoch": 5.7123451857770675,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 14298
    },
    {
      "epoch": 5.712744706352377,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 14299
    },
    {
      "epoch": 5.713144226927687,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 14300
    },
    {
      "epoch": 5.713543747502996,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 14301
    },
    {
      "epoch": 5.713943268078306,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 14302
    },
    {
      "epoch": 5.714342788653616,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 14303
    },
    {
      "epoch": 5.714742309228925,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 14304
    },
    {
      "epoch": 5.715141829804235,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 14305
    },
    {
      "epoch": 5.715541350379545,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 14306
    },
    {
      "epoch": 5.715940870954854,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 14307
    },
    {
      "epoch": 5.716340391530164,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 14308
    },
    {
      "epoch": 5.716739912105473,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 14309
    },
    {
      "epoch": 5.717139432680783,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 14310
    },
    {
      "epoch": 5.717538953256093,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 14311
    },
    {
      "epoch": 5.717938473831403,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 14312
    },
    {
      "epoch": 5.718337994406712,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 14313
    },
    {
      "epoch": 5.718737514982021,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 14314
    },
    {
      "epoch": 5.719137035557331,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 14315
    },
    {
      "epoch": 5.719536556132641,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 14316
    },
    {
      "epoch": 5.719936076707951,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 14317
    },
    {
      "epoch": 5.72033559728326,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 14318
    },
    {
      "epoch": 5.7207351178585695,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 14319
    },
    {
      "epoch": 5.721134638433879,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 14320
    },
    {
      "epoch": 5.721534159009189,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 14321
    },
    {
      "epoch": 5.721933679584499,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 14322
    },
    {
      "epoch": 5.722333200159808,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 14323
    },
    {
      "epoch": 5.7227327207351175,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 14324
    },
    {
      "epoch": 5.723132241310427,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 14325
    },
    {
      "epoch": 5.723531761885737,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 14326
    },
    {
      "epoch": 5.723931282461047,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 14327
    },
    {
      "epoch": 5.724330803036357,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 14328
    },
    {
      "epoch": 5.7247303236116664,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 14329
    },
    {
      "epoch": 5.725129844186975,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 14330
    },
    {
      "epoch": 5.725529364762285,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 14331
    },
    {
      "epoch": 5.725928885337595,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 14332
    },
    {
      "epoch": 5.726328405912905,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 14333
    },
    {
      "epoch": 5.7267279264882145,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 14334
    },
    {
      "epoch": 5.727127447063523,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 14335
    },
    {
      "epoch": 5.727526967638833,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 14336
    },
    {
      "epoch": 5.727926488214143,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 14337
    },
    {
      "epoch": 5.728326008789453,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 14338
    },
    {
      "epoch": 5.7287255293647625,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 14339
    },
    {
      "epoch": 5.729125049940071,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 14340
    },
    {
      "epoch": 5.729524570515381,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 14341
    },
    {
      "epoch": 5.729924091090691,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 14342
    },
    {
      "epoch": 5.730323611666001,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 14343
    },
    {
      "epoch": 5.730723132241311,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 14344
    },
    {
      "epoch": 5.73112265281662,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 14345
    },
    {
      "epoch": 5.73152217339193,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 14346
    },
    {
      "epoch": 5.731921693967239,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 14347
    },
    {
      "epoch": 5.732321214542549,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 14348
    },
    {
      "epoch": 5.732720735117859,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 14349
    },
    {
      "epoch": 5.733120255693168,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 14350
    },
    {
      "epoch": 5.733519776268478,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 14351
    },
    {
      "epoch": 5.733919296843787,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 14352
    },
    {
      "epoch": 5.734318817419097,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 14353
    },
    {
      "epoch": 5.734718337994407,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0561,
      "step": 14354
    },
    {
      "epoch": 5.7351178585697165,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 14355
    },
    {
      "epoch": 5.735517379145026,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 14356
    },
    {
      "epoch": 5.735916899720335,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 14357
    },
    {
      "epoch": 5.736316420295645,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 14358
    },
    {
      "epoch": 5.736715940870955,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 14359
    },
    {
      "epoch": 5.7371154614462645,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 14360
    },
    {
      "epoch": 5.737514982021574,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 14361
    },
    {
      "epoch": 5.737914502596884,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 14362
    },
    {
      "epoch": 5.738314023172194,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 14363
    },
    {
      "epoch": 5.738713543747503,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 14364
    },
    {
      "epoch": 5.7391130643228125,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 14365
    },
    {
      "epoch": 5.739512584898122,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 14366
    },
    {
      "epoch": 5.739912105473432,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 14367
    },
    {
      "epoch": 5.740311626048742,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 14368
    },
    {
      "epoch": 5.740711146624051,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 14369
    },
    {
      "epoch": 5.741110667199361,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 14370
    },
    {
      "epoch": 5.74151018777467,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 14371
    },
    {
      "epoch": 5.74190970834998,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 14372
    },
    {
      "epoch": 5.74230922892529,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 14373
    },
    {
      "epoch": 5.742708749500599,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 14374
    },
    {
      "epoch": 5.743108270075909,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 14375
    },
    {
      "epoch": 5.743507790651218,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 14376
    },
    {
      "epoch": 5.743907311226528,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 14377
    },
    {
      "epoch": 5.744306831801838,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 14378
    },
    {
      "epoch": 5.744706352377148,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 14379
    },
    {
      "epoch": 5.745105872952457,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 14380
    },
    {
      "epoch": 5.7455053935277665,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 14381
    },
    {
      "epoch": 5.745904914103076,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 14382
    },
    {
      "epoch": 5.746304434678386,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 14383
    },
    {
      "epoch": 5.746703955253696,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 14384
    },
    {
      "epoch": 5.747103475829006,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 14385
    },
    {
      "epoch": 5.7475029964043145,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 14386
    },
    {
      "epoch": 5.747902516979624,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 14387
    },
    {
      "epoch": 5.748302037554934,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 14388
    },
    {
      "epoch": 5.748701558130244,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 14389
    },
    {
      "epoch": 5.749101078705554,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 14390
    },
    {
      "epoch": 5.7495005992808625,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 14391
    },
    {
      "epoch": 5.749900119856172,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 14392
    },
    {
      "epoch": 5.750299640431482,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 14393
    },
    {
      "epoch": 5.750699161006792,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 14394
    },
    {
      "epoch": 5.751098681582102,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 14395
    },
    {
      "epoch": 5.7514982021574115,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 14396
    },
    {
      "epoch": 5.75189772273272,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 14397
    },
    {
      "epoch": 5.75229724330803,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 14398
    },
    {
      "epoch": 5.75269676388334,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 14399
    },
    {
      "epoch": 5.75309628445865,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 14400
    },
    {
      "epoch": 5.7534958050339595,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 14401
    },
    {
      "epoch": 5.753895325609269,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 14402
    },
    {
      "epoch": 5.754294846184578,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 14403
    },
    {
      "epoch": 5.754694366759888,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 14404
    },
    {
      "epoch": 5.755093887335198,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 14405
    },
    {
      "epoch": 5.755493407910508,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 14406
    },
    {
      "epoch": 5.755892928485817,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 14407
    },
    {
      "epoch": 5.756292449061126,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 14408
    },
    {
      "epoch": 5.756691969636436,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 14409
    },
    {
      "epoch": 5.757091490211746,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 14410
    },
    {
      "epoch": 5.757491010787056,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 14411
    },
    {
      "epoch": 5.757890531362365,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 14412
    },
    {
      "epoch": 5.758290051937674,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 14413
    },
    {
      "epoch": 5.758689572512984,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 14414
    },
    {
      "epoch": 5.759089093088294,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 14415
    },
    {
      "epoch": 5.759488613663604,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 14416
    },
    {
      "epoch": 5.759888134238913,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 14417
    },
    {
      "epoch": 5.760287654814223,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 14418
    },
    {
      "epoch": 5.760687175389533,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 14419
    },
    {
      "epoch": 5.761086695964842,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 14420
    },
    {
      "epoch": 5.761486216540152,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 14421
    },
    {
      "epoch": 5.7618857371154615,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 14422
    },
    {
      "epoch": 5.762285257690771,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 14423
    },
    {
      "epoch": 5.762684778266081,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 14424
    },
    {
      "epoch": 5.76308429884139,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 14425
    },
    {
      "epoch": 5.7634838194167,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 14426
    },
    {
      "epoch": 5.7638833399920095,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 14427
    },
    {
      "epoch": 5.764282860567319,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 14428
    },
    {
      "epoch": 5.764682381142629,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 14429
    },
    {
      "epoch": 5.765081901717938,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 14430
    },
    {
      "epoch": 5.765481422293248,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 14431
    },
    {
      "epoch": 5.765880942868558,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 14432
    },
    {
      "epoch": 5.766280463443867,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 14433
    },
    {
      "epoch": 5.766679984019177,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 14434
    },
    {
      "epoch": 5.767079504594487,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 14435
    },
    {
      "epoch": 5.767479025169797,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 14436
    },
    {
      "epoch": 5.767878545745106,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 14437
    },
    {
      "epoch": 5.768278066320415,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 14438
    },
    {
      "epoch": 5.768677586895725,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 14439
    },
    {
      "epoch": 5.769077107471035,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 14440
    },
    {
      "epoch": 5.769476628046345,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 14441
    },
    {
      "epoch": 5.769876148621654,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 14442
    },
    {
      "epoch": 5.770275669196963,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 14443
    },
    {
      "epoch": 5.770675189772273,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 14444
    },
    {
      "epoch": 5.771074710347583,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 14445
    },
    {
      "epoch": 5.771474230922893,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 14446
    },
    {
      "epoch": 5.771873751498202,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 14447
    },
    {
      "epoch": 5.7722732720735115,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 14448
    },
    {
      "epoch": 5.772672792648821,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 14449
    },
    {
      "epoch": 5.773072313224131,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 14450
    },
    {
      "epoch": 5.773471833799441,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 14451
    },
    {
      "epoch": 5.773871354374751,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 14452
    },
    {
      "epoch": 5.77427087495006,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 14453
    },
    {
      "epoch": 5.774670395525369,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 14454
    },
    {
      "epoch": 5.775069916100679,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 14455
    },
    {
      "epoch": 5.775469436675989,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 14456
    },
    {
      "epoch": 5.775868957251299,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 14457
    },
    {
      "epoch": 5.7762684778266085,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 14458
    },
    {
      "epoch": 5.776667998401917,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 14459
    },
    {
      "epoch": 5.777067518977227,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 14460
    },
    {
      "epoch": 5.777467039552537,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 14461
    },
    {
      "epoch": 5.777866560127847,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 14462
    },
    {
      "epoch": 5.7782660807031565,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 14463
    },
    {
      "epoch": 5.778665601278465,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 14464
    },
    {
      "epoch": 5.779065121853775,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 14465
    },
    {
      "epoch": 5.779464642429085,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 14466
    },
    {
      "epoch": 5.779864163004395,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 14467
    },
    {
      "epoch": 5.7802636835797045,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 14468
    },
    {
      "epoch": 5.780663204155014,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 14469
    },
    {
      "epoch": 5.781062724730323,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 14470
    },
    {
      "epoch": 5.781462245305633,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 14471
    },
    {
      "epoch": 5.781861765880943,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 14472
    },
    {
      "epoch": 5.782261286456253,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 14473
    },
    {
      "epoch": 5.782660807031562,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 14474
    },
    {
      "epoch": 5.783060327606872,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 14475
    },
    {
      "epoch": 5.783459848182181,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 14476
    },
    {
      "epoch": 5.783859368757491,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 14477
    },
    {
      "epoch": 5.784258889332801,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 14478
    },
    {
      "epoch": 5.78465840990811,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 14479
    },
    {
      "epoch": 5.78505793048342,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 14480
    },
    {
      "epoch": 5.785457451058729,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 14481
    },
    {
      "epoch": 5.785856971634039,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 14482
    },
    {
      "epoch": 5.786256492209349,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 14483
    },
    {
      "epoch": 5.7866560127846585,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 14484
    },
    {
      "epoch": 5.787055533359968,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 14485
    },
    {
      "epoch": 5.787455053935278,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 14486
    },
    {
      "epoch": 5.787854574510587,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 14487
    },
    {
      "epoch": 5.788254095085897,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 14488
    },
    {
      "epoch": 5.7886536156612065,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 14489
    },
    {
      "epoch": 5.789053136236516,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 14490
    },
    {
      "epoch": 5.789452656811826,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 14491
    },
    {
      "epoch": 5.789852177387136,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 14492
    },
    {
      "epoch": 5.790251697962445,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 14493
    },
    {
      "epoch": 5.7906512185377546,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 14494
    },
    {
      "epoch": 5.791050739113064,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 14495
    },
    {
      "epoch": 5.791450259688374,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 14496
    },
    {
      "epoch": 5.791849780263684,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 14497
    },
    {
      "epoch": 5.792249300838993,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 14498
    },
    {
      "epoch": 5.792648821414303,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 14499
    },
    {
      "epoch": 5.793048341989612,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 14500
    },
    {
      "epoch": 5.793447862564922,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 14501
    },
    {
      "epoch": 5.793847383140232,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 14502
    },
    {
      "epoch": 5.794246903715541,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 14503
    },
    {
      "epoch": 5.794646424290851,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 14504
    },
    {
      "epoch": 5.79504594486616,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 14505
    },
    {
      "epoch": 5.79544546544147,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 14506
    },
    {
      "epoch": 5.79584498601678,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 14507
    },
    {
      "epoch": 5.79624450659209,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 14508
    },
    {
      "epoch": 5.7966440271674,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 14509
    },
    {
      "epoch": 5.7970435477427085,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 14510
    },
    {
      "epoch": 5.797443068318018,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 14511
    },
    {
      "epoch": 5.797842588893328,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 14512
    },
    {
      "epoch": 5.798242109468638,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 14513
    },
    {
      "epoch": 5.798641630043948,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 14514
    },
    {
      "epoch": 5.7990411506192565,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 14515
    },
    {
      "epoch": 5.799440671194566,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 14516
    },
    {
      "epoch": 5.799840191769876,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 14517
    },
    {
      "epoch": 5.800239712345186,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 14518
    },
    {
      "epoch": 5.800639232920496,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 14519
    },
    {
      "epoch": 5.801038753495805,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 14520
    },
    {
      "epoch": 5.801438274071114,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 14521
    },
    {
      "epoch": 5.801837794646424,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 14522
    },
    {
      "epoch": 5.802237315221734,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 14523
    },
    {
      "epoch": 5.802636835797044,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 14524
    },
    {
      "epoch": 5.8030363563723535,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 14525
    },
    {
      "epoch": 5.803435876947663,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 14526
    },
    {
      "epoch": 5.803835397522972,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 14527
    },
    {
      "epoch": 5.804234918098282,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 14528
    },
    {
      "epoch": 5.804634438673592,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 14529
    },
    {
      "epoch": 5.8050339592489015,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 14530
    },
    {
      "epoch": 5.805433479824211,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 14531
    },
    {
      "epoch": 5.80583300039952,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 14532
    },
    {
      "epoch": 5.80623252097483,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 14533
    },
    {
      "epoch": 5.80663204155014,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 14534
    },
    {
      "epoch": 5.80703156212545,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 14535
    },
    {
      "epoch": 5.807431082700759,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 14536
    },
    {
      "epoch": 5.807830603276068,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 14537
    },
    {
      "epoch": 5.808230123851378,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 14538
    },
    {
      "epoch": 5.808629644426688,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 14539
    },
    {
      "epoch": 5.809029165001998,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 14540
    },
    {
      "epoch": 5.809428685577307,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 14541
    },
    {
      "epoch": 5.809828206152617,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 14542
    },
    {
      "epoch": 5.810227726727927,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 14543
    },
    {
      "epoch": 5.810627247303236,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 14544
    },
    {
      "epoch": 5.811026767878546,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 14545
    },
    {
      "epoch": 5.8114262884538554,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 14546
    },
    {
      "epoch": 5.811825809029165,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 14547
    },
    {
      "epoch": 5.812225329604475,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 14548
    },
    {
      "epoch": 5.812624850179784,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 14549
    },
    {
      "epoch": 5.813024370755094,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 14550
    },
    {
      "epoch": 5.8134238913304035,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 14551
    },
    {
      "epoch": 5.813823411905713,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 14552
    },
    {
      "epoch": 5.814222932481023,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 14553
    },
    {
      "epoch": 5.814622453056332,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 14554
    },
    {
      "epoch": 5.815021973631642,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 14555
    },
    {
      "epoch": 5.8154214942069515,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 14556
    },
    {
      "epoch": 5.815821014782261,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 14557
    },
    {
      "epoch": 5.816220535357571,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 14558
    },
    {
      "epoch": 5.816620055932881,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 14559
    },
    {
      "epoch": 5.81701957650819,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 14560
    },
    {
      "epoch": 5.8174190970835,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 14561
    },
    {
      "epoch": 5.817818617658809,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 14562
    },
    {
      "epoch": 5.818218138234119,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 14563
    },
    {
      "epoch": 5.818617658809429,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 14564
    },
    {
      "epoch": 5.819017179384739,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 14565
    },
    {
      "epoch": 5.819416699960048,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 14566
    },
    {
      "epoch": 5.819816220535357,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 14567
    },
    {
      "epoch": 5.820215741110667,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 14568
    },
    {
      "epoch": 5.820615261685977,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 14569
    },
    {
      "epoch": 5.821014782261287,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 14570
    },
    {
      "epoch": 5.821414302836596,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 14571
    },
    {
      "epoch": 5.8218138234119055,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 14572
    },
    {
      "epoch": 5.822213343987215,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 14573
    },
    {
      "epoch": 5.822612864562525,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 14574
    },
    {
      "epoch": 5.823012385137835,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 14575
    },
    {
      "epoch": 5.823411905713145,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 14576
    },
    {
      "epoch": 5.8238114262884535,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 14577
    },
    {
      "epoch": 5.824210946863763,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 14578
    },
    {
      "epoch": 5.824610467439073,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 14579
    },
    {
      "epoch": 5.825009988014383,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 14580
    },
    {
      "epoch": 5.825409508589693,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 14581
    },
    {
      "epoch": 5.825809029165002,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 14582
    },
    {
      "epoch": 5.826208549740311,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 14583
    },
    {
      "epoch": 5.826608070315621,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 14584
    },
    {
      "epoch": 5.827007590890931,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 14585
    },
    {
      "epoch": 5.827407111466241,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 14586
    },
    {
      "epoch": 5.8278066320415505,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 14587
    },
    {
      "epoch": 5.828206152616859,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 14588
    },
    {
      "epoch": 5.828605673192169,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 14589
    },
    {
      "epoch": 5.829005193767479,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 14590
    },
    {
      "epoch": 5.829404714342789,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 14591
    },
    {
      "epoch": 5.8298042349180985,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 14592
    },
    {
      "epoch": 5.830203755493407,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 14593
    },
    {
      "epoch": 5.830603276068717,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 14594
    },
    {
      "epoch": 5.831002796644027,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 14595
    },
    {
      "epoch": 5.831402317219337,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 14596
    },
    {
      "epoch": 5.831801837794647,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 14597
    },
    {
      "epoch": 5.832201358369956,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 14598
    },
    {
      "epoch": 5.832600878945266,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 14599
    },
    {
      "epoch": 5.833000399520575,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 14600
    },
    {
      "epoch": 5.833399920095885,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 14601
    },
    {
      "epoch": 5.833799440671195,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 14602
    },
    {
      "epoch": 5.834198961246504,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 14603
    },
    {
      "epoch": 5.834598481821814,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 14604
    },
    {
      "epoch": 5.834998002397123,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 14605
    },
    {
      "epoch": 5.835397522972433,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 14606
    },
    {
      "epoch": 5.835797043547743,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 14607
    },
    {
      "epoch": 5.836196564123052,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 14608
    },
    {
      "epoch": 5.836596084698362,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 14609
    },
    {
      "epoch": 5.836995605273671,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 14610
    },
    {
      "epoch": 5.837395125848981,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 14611
    },
    {
      "epoch": 5.837794646424291,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 14612
    },
    {
      "epoch": 5.8381941669996005,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 14613
    },
    {
      "epoch": 5.83859368757491,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 14614
    },
    {
      "epoch": 5.83899320815022,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 14615
    },
    {
      "epoch": 5.83939272872553,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 14616
    },
    {
      "epoch": 5.839792249300839,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 14617
    },
    {
      "epoch": 5.8401917698761485,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 14618
    },
    {
      "epoch": 5.840591290451458,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 14619
    },
    {
      "epoch": 5.840990811026768,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 14620
    },
    {
      "epoch": 5.841390331602078,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 14621
    },
    {
      "epoch": 5.841789852177387,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 14622
    },
    {
      "epoch": 5.842189372752697,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 14623
    },
    {
      "epoch": 5.842588893328006,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 14624
    },
    {
      "epoch": 5.842988413903316,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 14625
    },
    {
      "epoch": 5.843387934478626,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 14626
    },
    {
      "epoch": 5.843787455053935,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 14627
    },
    {
      "epoch": 5.844186975629245,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 14628
    },
    {
      "epoch": 5.844586496204554,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 14629
    },
    {
      "epoch": 5.844986016779864,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 14630
    },
    {
      "epoch": 5.845385537355174,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 14631
    },
    {
      "epoch": 5.845785057930484,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 14632
    },
    {
      "epoch": 5.8461845785057935,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 14633
    },
    {
      "epoch": 5.846584099081102,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 14634
    },
    {
      "epoch": 5.846983619656412,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 14635
    },
    {
      "epoch": 5.847383140231722,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 14636
    },
    {
      "epoch": 5.847782660807032,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 14637
    },
    {
      "epoch": 5.848182181382342,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 14638
    },
    {
      "epoch": 5.8485817019576505,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 14639
    },
    {
      "epoch": 5.84898122253296,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 14640
    },
    {
      "epoch": 5.84938074310827,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 14641
    },
    {
      "epoch": 5.84978026368358,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 14642
    },
    {
      "epoch": 5.85017978425889,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 14643
    },
    {
      "epoch": 5.8505793048341985,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 14644
    },
    {
      "epoch": 5.850978825409508,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 14645
    },
    {
      "epoch": 5.851378345984818,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 14646
    },
    {
      "epoch": 5.851777866560128,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 14647
    },
    {
      "epoch": 5.852177387135438,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 14648
    },
    {
      "epoch": 5.8525769077107475,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 14649
    },
    {
      "epoch": 5.852976428286056,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 14650
    },
    {
      "epoch": 5.853375948861366,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 14651
    },
    {
      "epoch": 5.853775469436676,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 14652
    },
    {
      "epoch": 5.854174990011986,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 14653
    },
    {
      "epoch": 5.8545745105872955,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 14654
    },
    {
      "epoch": 5.854974031162605,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 14655
    },
    {
      "epoch": 5.855373551737914,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 14656
    },
    {
      "epoch": 5.855773072313224,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 14657
    },
    {
      "epoch": 5.856172592888534,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 14658
    },
    {
      "epoch": 5.8565721134638435,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 14659
    },
    {
      "epoch": 5.856971634039153,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 14660
    },
    {
      "epoch": 5.857371154614462,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 14661
    },
    {
      "epoch": 5.857770675189772,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 14662
    },
    {
      "epoch": 5.858170195765082,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 14663
    },
    {
      "epoch": 5.858569716340392,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 14664
    },
    {
      "epoch": 5.858969236915701,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 14665
    },
    {
      "epoch": 5.859368757491011,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 14666
    },
    {
      "epoch": 5.85976827806632,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 14667
    },
    {
      "epoch": 5.86016779864163,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 14668
    },
    {
      "epoch": 5.86056731921694,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 14669
    },
    {
      "epoch": 5.860966839792249,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 14670
    },
    {
      "epoch": 5.861366360367559,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 14671
    },
    {
      "epoch": 5.861765880942869,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 14672
    },
    {
      "epoch": 5.862165401518178,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 14673
    },
    {
      "epoch": 5.862564922093488,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 14674
    },
    {
      "epoch": 5.8629644426687975,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 14675
    },
    {
      "epoch": 5.863363963244107,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 14676
    },
    {
      "epoch": 5.863763483819417,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 14677
    },
    {
      "epoch": 5.864163004394726,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 14678
    },
    {
      "epoch": 5.864562524970036,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 14679
    },
    {
      "epoch": 5.8649620455453455,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 14680
    },
    {
      "epoch": 5.865361566120655,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 14681
    },
    {
      "epoch": 5.865761086695965,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 14682
    },
    {
      "epoch": 5.866160607271274,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 14683
    },
    {
      "epoch": 5.866560127846584,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 14684
    },
    {
      "epoch": 5.8669596484218935,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 14685
    },
    {
      "epoch": 5.867359168997203,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 14686
    },
    {
      "epoch": 5.867758689572513,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 14687
    },
    {
      "epoch": 5.868158210147823,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 14688
    },
    {
      "epoch": 5.868557730723133,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 14689
    },
    {
      "epoch": 5.868957251298442,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 14690
    },
    {
      "epoch": 5.869356771873751,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 14691
    },
    {
      "epoch": 5.869756292449061,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 14692
    },
    {
      "epoch": 5.870155813024371,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 14693
    },
    {
      "epoch": 5.870555333599681,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 14694
    },
    {
      "epoch": 5.87095485417499,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 14695
    },
    {
      "epoch": 5.871354374750299,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 14696
    },
    {
      "epoch": 5.871753895325609,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 14697
    },
    {
      "epoch": 5.872153415900919,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 14698
    },
    {
      "epoch": 5.872552936476229,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 14699
    },
    {
      "epoch": 5.872952457051538,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 14700
    },
    {
      "epoch": 5.8733519776268475,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 14701
    },
    {
      "epoch": 5.873751498202157,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 14702
    },
    {
      "epoch": 5.874151018777467,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 14703
    },
    {
      "epoch": 5.874550539352777,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 14704
    },
    {
      "epoch": 5.874950059928087,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 14705
    },
    {
      "epoch": 5.875349580503396,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0551,
      "step": 14706
    },
    {
      "epoch": 5.875749101078705,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 14707
    },
    {
      "epoch": 5.876148621654015,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 14708
    },
    {
      "epoch": 5.876548142229325,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 14709
    },
    {
      "epoch": 5.876947662804635,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 14710
    },
    {
      "epoch": 5.877347183379944,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 14711
    },
    {
      "epoch": 5.877746703955253,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 14712
    },
    {
      "epoch": 5.878146224530563,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 14713
    },
    {
      "epoch": 5.878545745105873,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 14714
    },
    {
      "epoch": 5.878945265681183,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 14715
    },
    {
      "epoch": 5.8793447862564925,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 14716
    },
    {
      "epoch": 5.879744306831801,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 14717
    },
    {
      "epoch": 5.880143827407111,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 14718
    },
    {
      "epoch": 5.880543347982421,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 14719
    },
    {
      "epoch": 5.880942868557731,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 14720
    },
    {
      "epoch": 5.8813423891330405,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 14721
    },
    {
      "epoch": 5.88174190970835,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 14722
    },
    {
      "epoch": 5.88214143028366,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 14723
    },
    {
      "epoch": 5.882540950858969,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 14724
    },
    {
      "epoch": 5.882940471434279,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 14725
    },
    {
      "epoch": 5.883339992009589,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 14726
    },
    {
      "epoch": 5.883739512584898,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 14727
    },
    {
      "epoch": 5.884139033160208,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 14728
    },
    {
      "epoch": 5.884538553735517,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 14729
    },
    {
      "epoch": 5.884938074310827,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 14730
    },
    {
      "epoch": 5.885337594886137,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 14731
    },
    {
      "epoch": 5.885737115461446,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 14732
    },
    {
      "epoch": 5.886136636036756,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 14733
    },
    {
      "epoch": 5.886536156612065,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 14734
    },
    {
      "epoch": 5.886935677187375,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 14735
    },
    {
      "epoch": 5.887335197762685,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 14736
    },
    {
      "epoch": 5.887734718337994,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 14737
    },
    {
      "epoch": 5.888134238913304,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 14738
    },
    {
      "epoch": 5.888533759488614,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 14739
    },
    {
      "epoch": 5.888933280063923,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 14740
    },
    {
      "epoch": 5.889332800639233,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0581,
      "step": 14741
    },
    {
      "epoch": 5.8897323212145425,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 14742
    },
    {
      "epoch": 5.890131841789852,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 14743
    },
    {
      "epoch": 5.890531362365162,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 14744
    },
    {
      "epoch": 5.890930882940472,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 14745
    },
    {
      "epoch": 5.891330403515781,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 14746
    },
    {
      "epoch": 5.8917299240910905,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 14747
    },
    {
      "epoch": 5.8921294446664,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 14748
    },
    {
      "epoch": 5.89252896524171,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 14749
    },
    {
      "epoch": 5.89292848581702,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 14750
    },
    {
      "epoch": 5.893328006392329,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 14751
    },
    {
      "epoch": 5.893727526967639,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 14752
    },
    {
      "epoch": 5.894127047542948,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 14753
    },
    {
      "epoch": 5.894526568118258,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 14754
    },
    {
      "epoch": 5.894926088693568,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 14755
    },
    {
      "epoch": 5.895325609268878,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 14756
    },
    {
      "epoch": 5.895725129844187,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 14757
    },
    {
      "epoch": 5.896124650419496,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 14758
    },
    {
      "epoch": 5.896524170994806,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 14759
    },
    {
      "epoch": 5.896923691570116,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 14760
    },
    {
      "epoch": 5.897323212145426,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 14761
    },
    {
      "epoch": 5.8977227327207356,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 14762
    },
    {
      "epoch": 5.8981222532960444,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 14763
    },
    {
      "epoch": 5.898521773871354,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 14764
    },
    {
      "epoch": 5.898921294446664,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 14765
    },
    {
      "epoch": 5.899320815021974,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 14766
    },
    {
      "epoch": 5.899720335597284,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 14767
    },
    {
      "epoch": 5.9001198561725925,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 14768
    },
    {
      "epoch": 5.900519376747902,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 14769
    },
    {
      "epoch": 5.900918897323212,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 14770
    },
    {
      "epoch": 5.901318417898522,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 14771
    },
    {
      "epoch": 5.901717938473832,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 14772
    },
    {
      "epoch": 5.9021174590491405,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 14773
    },
    {
      "epoch": 5.90251697962445,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 14774
    },
    {
      "epoch": 5.90291650019976,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 14775
    },
    {
      "epoch": 5.90331602077507,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 14776
    },
    {
      "epoch": 5.90371554135038,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 14777
    },
    {
      "epoch": 5.9041150619256895,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 14778
    },
    {
      "epoch": 5.904514582500999,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 14779
    },
    {
      "epoch": 5.904914103076308,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 14780
    },
    {
      "epoch": 5.905313623651618,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 14781
    },
    {
      "epoch": 5.905713144226928,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 14782
    },
    {
      "epoch": 5.9061126648022375,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 14783
    },
    {
      "epoch": 5.906512185377547,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 14784
    },
    {
      "epoch": 5.906911705952856,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 14785
    },
    {
      "epoch": 5.907311226528166,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 14786
    },
    {
      "epoch": 5.907710747103476,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0562,
      "step": 14787
    },
    {
      "epoch": 5.9081102676787856,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 14788
    },
    {
      "epoch": 5.908509788254095,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 14789
    },
    {
      "epoch": 5.908909308829404,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 14790
    },
    {
      "epoch": 5.909308829404714,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 14791
    },
    {
      "epoch": 5.909708349980024,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 14792
    },
    {
      "epoch": 5.910107870555334,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 14793
    },
    {
      "epoch": 5.910507391130643,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 14794
    },
    {
      "epoch": 5.910906911705953,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 14795
    },
    {
      "epoch": 5.911306432281263,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0571,
      "step": 14796
    },
    {
      "epoch": 5.911705952856572,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 14797
    },
    {
      "epoch": 5.912105473431882,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 14798
    },
    {
      "epoch": 5.912504994007191,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 14799
    },
    {
      "epoch": 5.912904514582501,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 14800
    },
    {
      "epoch": 5.913304035157811,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 14801
    },
    {
      "epoch": 5.91370355573312,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 14802
    },
    {
      "epoch": 5.91410307630843,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 14803
    },
    {
      "epoch": 5.9145025968837395,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 14804
    },
    {
      "epoch": 5.914902117459049,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 14805
    },
    {
      "epoch": 5.915301638034359,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 14806
    },
    {
      "epoch": 5.915701158609668,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 14807
    },
    {
      "epoch": 5.916100679184978,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 14808
    },
    {
      "epoch": 5.9165001997602875,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 14809
    },
    {
      "epoch": 5.916899720335597,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 14810
    },
    {
      "epoch": 5.917299240910907,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 14811
    },
    {
      "epoch": 5.917698761486217,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 14812
    },
    {
      "epoch": 5.918098282061527,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 14813
    },
    {
      "epoch": 5.918497802636836,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 14814
    },
    {
      "epoch": 5.918897323212145,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 14815
    },
    {
      "epoch": 5.919296843787455,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 14816
    },
    {
      "epoch": 5.919696364362765,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 14817
    },
    {
      "epoch": 5.920095884938075,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 14818
    },
    {
      "epoch": 5.920495405513384,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 14819
    },
    {
      "epoch": 5.920894926088693,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 14820
    },
    {
      "epoch": 5.921294446664003,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 14821
    },
    {
      "epoch": 5.921693967239313,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 14822
    },
    {
      "epoch": 5.922093487814623,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 14823
    },
    {
      "epoch": 5.922493008389932,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 14824
    },
    {
      "epoch": 5.922892528965241,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 14825
    },
    {
      "epoch": 5.923292049540551,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 14826
    },
    {
      "epoch": 5.923691570115861,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 14827
    },
    {
      "epoch": 5.924091090691171,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 14828
    },
    {
      "epoch": 5.924490611266481,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 14829
    },
    {
      "epoch": 5.9248901318417895,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 14830
    },
    {
      "epoch": 5.925289652417099,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 14831
    },
    {
      "epoch": 5.925689172992409,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 14832
    },
    {
      "epoch": 5.926088693567719,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 14833
    },
    {
      "epoch": 5.926488214143029,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 14834
    },
    {
      "epoch": 5.926887734718338,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 14835
    },
    {
      "epoch": 5.927287255293647,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 14836
    },
    {
      "epoch": 5.927686775868957,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 14837
    },
    {
      "epoch": 5.928086296444267,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 14838
    },
    {
      "epoch": 5.928485817019577,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 14839
    },
    {
      "epoch": 5.9288853375948865,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 14840
    },
    {
      "epoch": 5.929284858170195,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 14841
    },
    {
      "epoch": 5.929684378745505,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 14842
    },
    {
      "epoch": 5.930083899320815,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 14843
    },
    {
      "epoch": 5.930483419896125,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 14844
    },
    {
      "epoch": 5.9308829404714345,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 14845
    },
    {
      "epoch": 5.931282461046744,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 14846
    },
    {
      "epoch": 5.931681981622053,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 14847
    },
    {
      "epoch": 5.932081502197363,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 14848
    },
    {
      "epoch": 5.932481022772673,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 14849
    },
    {
      "epoch": 5.9328805433479825,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 14850
    },
    {
      "epoch": 5.933280063923292,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 14851
    },
    {
      "epoch": 5.933679584498602,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 14852
    },
    {
      "epoch": 5.934079105073911,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 14853
    },
    {
      "epoch": 5.934478625649221,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 14854
    },
    {
      "epoch": 5.934878146224531,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 14855
    },
    {
      "epoch": 5.93527766679984,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 14856
    },
    {
      "epoch": 5.93567718737515,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 14857
    },
    {
      "epoch": 5.936076707950459,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 14858
    },
    {
      "epoch": 5.936476228525769,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 14859
    },
    {
      "epoch": 5.936875749101079,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 14860
    },
    {
      "epoch": 5.937275269676388,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 14861
    },
    {
      "epoch": 5.937674790251698,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 14862
    },
    {
      "epoch": 5.938074310827007,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 14863
    },
    {
      "epoch": 5.938473831402317,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 14864
    },
    {
      "epoch": 5.938873351977627,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 14865
    },
    {
      "epoch": 5.9392728725529365,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 14866
    },
    {
      "epoch": 5.939672393128246,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 14867
    },
    {
      "epoch": 5.940071913703556,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 14868
    },
    {
      "epoch": 5.940471434278866,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 14869
    },
    {
      "epoch": 5.940870954854175,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 14870
    },
    {
      "epoch": 5.9412704754294845,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 14871
    },
    {
      "epoch": 5.941669996004794,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 14872
    },
    {
      "epoch": 5.942069516580104,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 14873
    },
    {
      "epoch": 5.942469037155414,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 14874
    },
    {
      "epoch": 5.942868557730723,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 14875
    },
    {
      "epoch": 5.9432680783060325,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 14876
    },
    {
      "epoch": 5.943667598881342,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 14877
    },
    {
      "epoch": 5.944067119456652,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 14878
    },
    {
      "epoch": 5.944466640031962,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 14879
    },
    {
      "epoch": 5.944866160607271,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 14880
    },
    {
      "epoch": 5.945265681182581,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 14881
    },
    {
      "epoch": 5.94566520175789,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 14882
    },
    {
      "epoch": 5.9460647223332,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 14883
    },
    {
      "epoch": 5.94646424290851,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 14884
    },
    {
      "epoch": 5.94686376348382,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 14885
    },
    {
      "epoch": 5.9472632840591295,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 14886
    },
    {
      "epoch": 5.947662804634438,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 14887
    },
    {
      "epoch": 5.948062325209748,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 14888
    },
    {
      "epoch": 5.948461845785058,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 14889
    },
    {
      "epoch": 5.948861366360368,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 14890
    },
    {
      "epoch": 5.949260886935678,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 14891
    },
    {
      "epoch": 5.9496604075109865,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 14892
    },
    {
      "epoch": 5.950059928086296,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 14893
    },
    {
      "epoch": 5.950459448661606,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 14894
    },
    {
      "epoch": 5.950858969236916,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 14895
    },
    {
      "epoch": 5.951258489812226,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 14896
    },
    {
      "epoch": 5.9516580103875345,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 14897
    },
    {
      "epoch": 5.952057530962844,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 14898
    },
    {
      "epoch": 5.952457051538154,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 14899
    },
    {
      "epoch": 5.952856572113464,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 14900
    },
    {
      "epoch": 5.953256092688774,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 14901
    },
    {
      "epoch": 5.953655613264083,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 14902
    },
    {
      "epoch": 5.954055133839393,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 14903
    },
    {
      "epoch": 5.954454654414702,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 14904
    },
    {
      "epoch": 5.954854174990012,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 14905
    },
    {
      "epoch": 5.955253695565322,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 14906
    },
    {
      "epoch": 5.9556532161406315,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 14907
    },
    {
      "epoch": 5.956052736715941,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 14908
    },
    {
      "epoch": 5.95645225729125,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 14909
    },
    {
      "epoch": 5.95685177786656,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 14910
    },
    {
      "epoch": 5.95725129844187,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 14911
    },
    {
      "epoch": 5.9576508190171795,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 14912
    },
    {
      "epoch": 5.958050339592489,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 14913
    },
    {
      "epoch": 5.958449860167798,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 14914
    },
    {
      "epoch": 5.958849380743108,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 14915
    },
    {
      "epoch": 5.959248901318418,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 14916
    },
    {
      "epoch": 5.959648421893728,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 14917
    },
    {
      "epoch": 5.960047942469037,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 14918
    },
    {
      "epoch": 5.960447463044347,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 14919
    },
    {
      "epoch": 5.960846983619656,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 14920
    },
    {
      "epoch": 5.961246504194966,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 14921
    },
    {
      "epoch": 5.961646024770276,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 14922
    },
    {
      "epoch": 5.962045545345585,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 14923
    },
    {
      "epoch": 5.962445065920895,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 14924
    },
    {
      "epoch": 5.962844586496205,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 14925
    },
    {
      "epoch": 5.963244107071514,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 14926
    },
    {
      "epoch": 5.963643627646824,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 14927
    },
    {
      "epoch": 5.964043148222133,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 14928
    },
    {
      "epoch": 5.964442668797443,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 14929
    },
    {
      "epoch": 5.964842189372753,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 14930
    },
    {
      "epoch": 5.965241709948062,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 14931
    },
    {
      "epoch": 5.965641230523372,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 14932
    },
    {
      "epoch": 5.9660407510986815,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 14933
    },
    {
      "epoch": 5.966440271673991,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 14934
    },
    {
      "epoch": 5.966839792249301,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 14935
    },
    {
      "epoch": 5.967239312824611,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 14936
    },
    {
      "epoch": 5.96763883339992,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 14937
    },
    {
      "epoch": 5.9680383539752295,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 14938
    },
    {
      "epoch": 5.968437874550539,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 14939
    },
    {
      "epoch": 5.968837395125849,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 14940
    },
    {
      "epoch": 5.969236915701159,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 14941
    },
    {
      "epoch": 5.969636436276469,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 14942
    },
    {
      "epoch": 5.970035956851778,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 14943
    },
    {
      "epoch": 5.970435477427087,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 14944
    },
    {
      "epoch": 5.970834998002397,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 14945
    },
    {
      "epoch": 5.971234518577707,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 14946
    },
    {
      "epoch": 5.971634039153017,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 14947
    },
    {
      "epoch": 5.972033559728326,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 14948
    },
    {
      "epoch": 5.972433080303635,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 14949
    },
    {
      "epoch": 5.972832600878945,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 14950
    },
    {
      "epoch": 5.973232121454255,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 14951
    },
    {
      "epoch": 5.973631642029565,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 14952
    },
    {
      "epoch": 5.974031162604874,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 14953
    },
    {
      "epoch": 5.974430683180183,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 14954
    },
    {
      "epoch": 5.974830203755493,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 14955
    },
    {
      "epoch": 5.975229724330803,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 14956
    },
    {
      "epoch": 5.975629244906113,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 14957
    },
    {
      "epoch": 5.976028765481423,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 14958
    },
    {
      "epoch": 5.976428286056732,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 14959
    },
    {
      "epoch": 5.976827806632041,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 14960
    },
    {
      "epoch": 5.977227327207351,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 14961
    },
    {
      "epoch": 5.977626847782661,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 14962
    },
    {
      "epoch": 5.978026368357971,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 14963
    },
    {
      "epoch": 5.97842588893328,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 14964
    },
    {
      "epoch": 5.978825409508589,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 14965
    },
    {
      "epoch": 5.979224930083899,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 14966
    },
    {
      "epoch": 5.979624450659209,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 14967
    },
    {
      "epoch": 5.980023971234519,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 14968
    },
    {
      "epoch": 5.9804234918098285,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 14969
    },
    {
      "epoch": 5.980823012385137,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 14970
    },
    {
      "epoch": 5.981222532960447,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 14971
    },
    {
      "epoch": 5.981622053535757,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 14972
    },
    {
      "epoch": 5.982021574111067,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 14973
    },
    {
      "epoch": 5.9824210946863765,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 14974
    },
    {
      "epoch": 5.982820615261686,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 14975
    },
    {
      "epoch": 5.983220135836996,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 14976
    },
    {
      "epoch": 5.983619656412305,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 14977
    },
    {
      "epoch": 5.984019176987615,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 14978
    },
    {
      "epoch": 5.9844186975629245,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 14979
    },
    {
      "epoch": 5.984818218138234,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 14980
    },
    {
      "epoch": 5.985217738713544,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 14981
    },
    {
      "epoch": 5.985617259288853,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 14982
    },
    {
      "epoch": 5.986016779864163,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 14983
    },
    {
      "epoch": 5.986416300439473,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 14984
    },
    {
      "epoch": 5.986815821014782,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 14985
    },
    {
      "epoch": 5.987215341590092,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 14986
    },
    {
      "epoch": 5.987614862165401,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 14987
    },
    {
      "epoch": 5.988014382740711,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 14988
    },
    {
      "epoch": 5.988413903316021,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 14989
    },
    {
      "epoch": 5.98881342389133,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 14990
    },
    {
      "epoch": 5.98921294446664,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 14991
    },
    {
      "epoch": 5.98961246504195,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 14992
    },
    {
      "epoch": 5.99001198561726,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 14993
    },
    {
      "epoch": 5.990411506192569,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 14994
    },
    {
      "epoch": 5.9908110267678785,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 14995
    },
    {
      "epoch": 5.991210547343188,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 14996
    },
    {
      "epoch": 5.991610067918498,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 14997
    },
    {
      "epoch": 5.992009588493808,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 14998
    },
    {
      "epoch": 5.992409109069117,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 14999
    },
    {
      "epoch": 5.9928086296444265,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 15000
    },
    {
      "epoch": 5.993208150219736,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 15001
    },
    {
      "epoch": 5.993607670795046,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 15002
    },
    {
      "epoch": 5.994007191370356,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 15003
    },
    {
      "epoch": 5.994406711945665,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 15004
    },
    {
      "epoch": 5.9948062325209746,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 15005
    },
    {
      "epoch": 5.995205753096284,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 15006
    },
    {
      "epoch": 5.995605273671594,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 15007
    },
    {
      "epoch": 5.996004794246904,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 15008
    },
    {
      "epoch": 5.996404314822214,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 15009
    },
    {
      "epoch": 5.996803835397523,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 15010
    },
    {
      "epoch": 5.997203355972832,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 15011
    },
    {
      "epoch": 5.997602876548142,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 15012
    },
    {
      "epoch": 5.998002397123452,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 15013
    },
    {
      "epoch": 5.998401917698762,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 15014
    },
    {
      "epoch": 5.9988014382740715,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 15015
    },
    {
      "epoch": 5.99920095884938,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 15016
    },
    {
      "epoch": 5.99960047942469,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 15017
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 15018
    },
    {
      "epoch": 6.00039952057531,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 15019
    },
    {
      "epoch": 6.00079904115062,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 15020
    },
    {
      "epoch": 6.0011985617259285,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 15021
    },
    {
      "epoch": 6.001598082301238,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 15022
    },
    {
      "epoch": 6.001997602876548,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 15023
    },
    {
      "epoch": 6.002397123451858,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 15024
    },
    {
      "epoch": 6.002796644027168,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 15025
    },
    {
      "epoch": 6.003196164602477,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 15026
    },
    {
      "epoch": 6.003595685177786,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 15027
    },
    {
      "epoch": 6.003995205753096,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 15028
    },
    {
      "epoch": 6.004394726328406,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 15029
    },
    {
      "epoch": 6.004794246903716,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 15030
    },
    {
      "epoch": 6.0051937674790254,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 15031
    },
    {
      "epoch": 6.005593288054335,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 15032
    },
    {
      "epoch": 6.005992808629644,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 15033
    },
    {
      "epoch": 6.006392329204954,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 15034
    },
    {
      "epoch": 6.006791849780264,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 15035
    },
    {
      "epoch": 6.0071913703555735,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 15036
    },
    {
      "epoch": 6.007590890930883,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 15037
    },
    {
      "epoch": 6.007990411506192,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 15038
    },
    {
      "epoch": 6.008389932081502,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 15039
    },
    {
      "epoch": 6.008789452656812,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 15040
    },
    {
      "epoch": 6.0091889732321215,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 15041
    },
    {
      "epoch": 6.009588493807431,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 15042
    },
    {
      "epoch": 6.009988014382741,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 15043
    },
    {
      "epoch": 6.01038753495805,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 15044
    },
    {
      "epoch": 6.01078705553336,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 15045
    },
    {
      "epoch": 6.01118657610867,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 15046
    },
    {
      "epoch": 6.011586096683979,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 15047
    },
    {
      "epoch": 6.011985617259289,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 15048
    },
    {
      "epoch": 6.012385137834598,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 15049
    },
    {
      "epoch": 6.012784658409908,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 15050
    },
    {
      "epoch": 6.013184178985218,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 15051
    },
    {
      "epoch": 6.013583699560527,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 15052
    },
    {
      "epoch": 6.013983220135837,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 15053
    },
    {
      "epoch": 6.014382740711147,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 15054
    },
    {
      "epoch": 6.014782261286456,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 15055
    },
    {
      "epoch": 6.015181781861766,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 15056
    },
    {
      "epoch": 6.0155813024370755,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 15057
    },
    {
      "epoch": 6.015980823012385,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 15058
    },
    {
      "epoch": 6.016380343587695,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 15059
    },
    {
      "epoch": 6.016779864163005,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 15060
    },
    {
      "epoch": 6.017179384738314,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 15061
    },
    {
      "epoch": 6.0175789053136235,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.058,
      "step": 15062
    },
    {
      "epoch": 6.017978425888933,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 15063
    },
    {
      "epoch": 6.018377946464243,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 15064
    },
    {
      "epoch": 6.018777467039553,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 15065
    },
    {
      "epoch": 6.019176987614862,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 15066
    },
    {
      "epoch": 6.0195765081901715,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 15067
    },
    {
      "epoch": 6.019976028765481,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 15068
    },
    {
      "epoch": 6.020375549340791,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 15069
    },
    {
      "epoch": 6.020775069916101,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 15070
    },
    {
      "epoch": 6.021174590491411,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 15071
    },
    {
      "epoch": 6.02157411106672,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 15072
    },
    {
      "epoch": 6.021973631642029,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 15073
    },
    {
      "epoch": 6.022373152217339,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 15074
    },
    {
      "epoch": 6.022772672792649,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 15075
    },
    {
      "epoch": 6.023172193367959,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 15076
    },
    {
      "epoch": 6.0235717139432685,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 15077
    },
    {
      "epoch": 6.023971234518577,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 15078
    },
    {
      "epoch": 6.024370755093887,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 15079
    },
    {
      "epoch": 6.024770275669197,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 15080
    },
    {
      "epoch": 6.025169796244507,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 15081
    },
    {
      "epoch": 6.025569316819817,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 15082
    },
    {
      "epoch": 6.0259688373951255,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 15083
    },
    {
      "epoch": 6.026368357970435,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 15084
    },
    {
      "epoch": 6.026767878545745,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 15085
    },
    {
      "epoch": 6.027167399121055,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 15086
    },
    {
      "epoch": 6.027566919696365,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 15087
    },
    {
      "epoch": 6.027966440271674,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 15088
    },
    {
      "epoch": 6.028365960846983,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 15089
    },
    {
      "epoch": 6.028765481422293,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 15090
    },
    {
      "epoch": 6.029165001997603,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 15091
    },
    {
      "epoch": 6.029564522572913,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 15092
    },
    {
      "epoch": 6.029964043148222,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 15093
    },
    {
      "epoch": 6.030363563723531,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 15094
    },
    {
      "epoch": 6.030763084298841,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 15095
    },
    {
      "epoch": 6.031162604874151,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 15096
    },
    {
      "epoch": 6.031562125449461,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 15097
    },
    {
      "epoch": 6.0319616460247705,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 15098
    },
    {
      "epoch": 6.03236116660008,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 15099
    },
    {
      "epoch": 6.032760687175389,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 15100
    },
    {
      "epoch": 6.033160207750699,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 15101
    },
    {
      "epoch": 6.033559728326009,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 15102
    },
    {
      "epoch": 6.0339592489013185,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 15103
    },
    {
      "epoch": 6.034358769476628,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 15104
    },
    {
      "epoch": 6.034758290051938,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 15105
    },
    {
      "epoch": 6.035157810627247,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 15106
    },
    {
      "epoch": 6.035557331202557,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 15107
    },
    {
      "epoch": 6.035956851777867,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 15108
    },
    {
      "epoch": 6.036356372353176,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 15109
    },
    {
      "epoch": 6.036755892928486,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 15110
    },
    {
      "epoch": 6.037155413503795,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 15111
    },
    {
      "epoch": 6.037554934079105,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 15112
    },
    {
      "epoch": 6.037954454654415,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 15113
    },
    {
      "epoch": 6.038353975229724,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 15114
    },
    {
      "epoch": 6.038753495805034,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 15115
    },
    {
      "epoch": 6.039153016380344,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 15116
    },
    {
      "epoch": 6.039552536955653,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 15117
    },
    {
      "epoch": 6.039952057530963,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 15118
    },
    {
      "epoch": 6.040351578106272,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 15119
    },
    {
      "epoch": 6.040751098681582,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 15120
    },
    {
      "epoch": 6.041150619256892,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 15121
    },
    {
      "epoch": 6.041550139832202,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 15122
    },
    {
      "epoch": 6.041949660407511,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 15123
    },
    {
      "epoch": 6.0423491809828205,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 15124
    },
    {
      "epoch": 6.04274870155813,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 15125
    },
    {
      "epoch": 6.04314822213344,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 15126
    },
    {
      "epoch": 6.04354774270875,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 15127
    },
    {
      "epoch": 6.043947263284059,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 15128
    },
    {
      "epoch": 6.0443467838593685,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 15129
    },
    {
      "epoch": 6.044746304434678,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 15130
    },
    {
      "epoch": 6.045145825009988,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 15131
    },
    {
      "epoch": 6.045545345585298,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 15132
    },
    {
      "epoch": 6.045944866160608,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 15133
    },
    {
      "epoch": 6.046344386735917,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 15134
    },
    {
      "epoch": 6.046743907311226,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 15135
    },
    {
      "epoch": 6.047143427886536,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 15136
    },
    {
      "epoch": 6.047542948461846,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 15137
    },
    {
      "epoch": 6.047942469037156,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 15138
    },
    {
      "epoch": 6.048341989612465,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 15139
    },
    {
      "epoch": 6.048741510187774,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 15140
    },
    {
      "epoch": 6.049141030763084,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 15141
    },
    {
      "epoch": 6.049540551338394,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 15142
    },
    {
      "epoch": 6.049940071913704,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 15143
    },
    {
      "epoch": 6.0503395924890135,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 15144
    },
    {
      "epoch": 6.050739113064322,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 15145
    },
    {
      "epoch": 6.051138633639632,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 15146
    },
    {
      "epoch": 6.051538154214942,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 15147
    },
    {
      "epoch": 6.051937674790252,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 15148
    },
    {
      "epoch": 6.052337195365562,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 15149
    },
    {
      "epoch": 6.052736715940871,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 15150
    },
    {
      "epoch": 6.05313623651618,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 15151
    },
    {
      "epoch": 6.05353575709149,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 15152
    },
    {
      "epoch": 6.0539352776668,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 15153
    },
    {
      "epoch": 6.05433479824211,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 15154
    },
    {
      "epoch": 6.054734318817419,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 15155
    },
    {
      "epoch": 6.055133839392728,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 15156
    },
    {
      "epoch": 6.055533359968038,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 15157
    },
    {
      "epoch": 6.055932880543348,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 15158
    },
    {
      "epoch": 6.056332401118658,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 15159
    },
    {
      "epoch": 6.0567319216939675,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 15160
    },
    {
      "epoch": 6.057131442269277,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 15161
    },
    {
      "epoch": 6.057530962844586,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 15162
    },
    {
      "epoch": 6.057930483419896,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 15163
    },
    {
      "epoch": 6.058330003995206,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 15164
    },
    {
      "epoch": 6.0587295245705155,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 15165
    },
    {
      "epoch": 6.059129045145825,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 15166
    },
    {
      "epoch": 6.059528565721135,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 15167
    },
    {
      "epoch": 6.059928086296444,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 15168
    },
    {
      "epoch": 6.060327606871754,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 15169
    },
    {
      "epoch": 6.0607271274470635,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 15170
    },
    {
      "epoch": 6.061126648022373,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 15171
    },
    {
      "epoch": 6.061526168597683,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 15172
    },
    {
      "epoch": 6.061925689172992,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 15173
    },
    {
      "epoch": 6.062325209748302,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 15174
    },
    {
      "epoch": 6.062724730323612,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 15175
    },
    {
      "epoch": 6.063124250898921,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 15176
    },
    {
      "epoch": 6.063523771474231,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 15177
    },
    {
      "epoch": 6.063923292049541,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 15178
    },
    {
      "epoch": 6.06432281262485,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 15179
    },
    {
      "epoch": 6.06472233320016,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 15180
    },
    {
      "epoch": 6.065121853775469,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 15181
    },
    {
      "epoch": 6.065521374350779,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 15182
    },
    {
      "epoch": 6.065920894926089,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 15183
    },
    {
      "epoch": 6.066320415501398,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 15184
    },
    {
      "epoch": 6.066719936076708,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 15185
    },
    {
      "epoch": 6.0671194566520175,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 15186
    },
    {
      "epoch": 6.067518977227327,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 15187
    },
    {
      "epoch": 6.067918497802637,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 15188
    },
    {
      "epoch": 6.068318018377947,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 15189
    },
    {
      "epoch": 6.068717538953256,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 15190
    },
    {
      "epoch": 6.0691170595285655,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 15191
    },
    {
      "epoch": 6.069516580103875,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 15192
    },
    {
      "epoch": 6.069916100679185,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 15193
    },
    {
      "epoch": 6.070315621254495,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 15194
    },
    {
      "epoch": 6.070715141829805,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 15195
    },
    {
      "epoch": 6.0711146624051135,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 15196
    },
    {
      "epoch": 6.071514182980423,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 15197
    },
    {
      "epoch": 6.071913703555733,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 15198
    },
    {
      "epoch": 6.072313224131043,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 15199
    },
    {
      "epoch": 6.072712744706353,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 15200
    },
    {
      "epoch": 6.073112265281662,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 15201
    },
    {
      "epoch": 6.073511785856971,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 15202
    },
    {
      "epoch": 6.073911306432281,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 15203
    },
    {
      "epoch": 6.074310827007591,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 15204
    },
    {
      "epoch": 6.074710347582901,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 15205
    },
    {
      "epoch": 6.0751098681582105,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 15206
    },
    {
      "epoch": 6.075509388733519,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 15207
    },
    {
      "epoch": 6.075908909308829,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 15208
    },
    {
      "epoch": 6.076308429884139,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 15209
    },
    {
      "epoch": 6.076707950459449,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 15210
    },
    {
      "epoch": 6.077107471034759,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 15211
    },
    {
      "epoch": 6.077506991610068,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 15212
    },
    {
      "epoch": 6.077906512185377,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 15213
    },
    {
      "epoch": 6.078306032760687,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 15214
    },
    {
      "epoch": 6.078705553335997,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 15215
    },
    {
      "epoch": 6.079105073911307,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 15216
    },
    {
      "epoch": 6.079504594486616,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 15217
    },
    {
      "epoch": 6.079904115061925,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 15218
    },
    {
      "epoch": 6.080303635637235,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 15219
    },
    {
      "epoch": 6.080703156212545,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 15220
    },
    {
      "epoch": 6.081102676787855,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 15221
    },
    {
      "epoch": 6.081502197363164,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 15222
    },
    {
      "epoch": 6.081901717938474,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 15223
    },
    {
      "epoch": 6.082301238513783,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 15224
    },
    {
      "epoch": 6.082700759089093,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 15225
    },
    {
      "epoch": 6.083100279664403,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 15226
    },
    {
      "epoch": 6.0834998002397125,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 15227
    },
    {
      "epoch": 6.083899320815022,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 15228
    },
    {
      "epoch": 6.084298841390331,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 15229
    },
    {
      "epoch": 6.084698361965641,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 15230
    },
    {
      "epoch": 6.085097882540951,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 15231
    },
    {
      "epoch": 6.0854974031162605,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 15232
    },
    {
      "epoch": 6.08589692369157,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 15233
    },
    {
      "epoch": 6.08629644426688,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 15234
    },
    {
      "epoch": 6.086695964842189,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 15235
    },
    {
      "epoch": 6.087095485417499,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 15236
    },
    {
      "epoch": 6.087495005992809,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 15237
    },
    {
      "epoch": 6.087894526568118,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 15238
    },
    {
      "epoch": 6.088294047143428,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 15239
    },
    {
      "epoch": 6.088693567718738,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 15240
    },
    {
      "epoch": 6.089093088294047,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 15241
    },
    {
      "epoch": 6.089492608869357,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 15242
    },
    {
      "epoch": 6.089892129444666,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 15243
    },
    {
      "epoch": 6.090291650019976,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 15244
    },
    {
      "epoch": 6.090691170595286,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 15245
    },
    {
      "epoch": 6.091090691170595,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 15246
    },
    {
      "epoch": 6.091490211745905,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 15247
    },
    {
      "epoch": 6.0918897323212144,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 15248
    },
    {
      "epoch": 6.092289252896524,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 15249
    },
    {
      "epoch": 6.092688773471834,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 15250
    },
    {
      "epoch": 6.093088294047144,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 15251
    },
    {
      "epoch": 6.093487814622453,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 15252
    },
    {
      "epoch": 6.0938873351977625,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 15253
    },
    {
      "epoch": 6.094286855773072,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 15254
    },
    {
      "epoch": 6.094686376348382,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 15255
    },
    {
      "epoch": 6.095085896923692,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 15256
    },
    {
      "epoch": 6.095485417499002,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 15257
    },
    {
      "epoch": 6.0958849380743105,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 15258
    },
    {
      "epoch": 6.09628445864962,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 15259
    },
    {
      "epoch": 6.09668397922493,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 15260
    },
    {
      "epoch": 6.09708349980024,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 15261
    },
    {
      "epoch": 6.09748302037555,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 15262
    },
    {
      "epoch": 6.097882540950859,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 15263
    },
    {
      "epoch": 6.098282061526168,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 15264
    },
    {
      "epoch": 6.098681582101478,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 15265
    },
    {
      "epoch": 6.099081102676788,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 15266
    },
    {
      "epoch": 6.099480623252098,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 15267
    },
    {
      "epoch": 6.0998801438274075,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 15268
    },
    {
      "epoch": 6.100279664402716,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 15269
    },
    {
      "epoch": 6.100679184978026,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 15270
    },
    {
      "epoch": 6.101078705553336,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 15271
    },
    {
      "epoch": 6.101478226128646,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 15272
    },
    {
      "epoch": 6.1018777467039556,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 15273
    },
    {
      "epoch": 6.1022772672792644,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 15274
    },
    {
      "epoch": 6.102676787854574,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 15275
    },
    {
      "epoch": 6.103076308429884,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 15276
    },
    {
      "epoch": 6.103475829005194,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 15277
    },
    {
      "epoch": 6.103875349580504,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 15278
    },
    {
      "epoch": 6.104274870155813,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 15279
    },
    {
      "epoch": 6.104674390731122,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 15280
    },
    {
      "epoch": 6.105073911306432,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 15281
    },
    {
      "epoch": 6.105473431881742,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 15282
    },
    {
      "epoch": 6.105872952457052,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 15283
    },
    {
      "epoch": 6.106272473032361,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 15284
    },
    {
      "epoch": 6.106671993607671,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 15285
    },
    {
      "epoch": 6.10707151418298,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 15286
    },
    {
      "epoch": 6.10747103475829,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 15287
    },
    {
      "epoch": 6.1078705553336,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 15288
    },
    {
      "epoch": 6.1082700759089095,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 15289
    },
    {
      "epoch": 6.108669596484219,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 15290
    },
    {
      "epoch": 6.109069117059528,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 15291
    },
    {
      "epoch": 6.109468637634838,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 15292
    },
    {
      "epoch": 6.109868158210148,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 15293
    },
    {
      "epoch": 6.1102676787854575,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 15294
    },
    {
      "epoch": 6.110667199360767,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 15295
    },
    {
      "epoch": 6.111066719936077,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 15296
    },
    {
      "epoch": 6.111466240511386,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 15297
    },
    {
      "epoch": 6.111865761086696,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 15298
    },
    {
      "epoch": 6.112265281662006,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 15299
    },
    {
      "epoch": 6.112664802237315,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 15300
    },
    {
      "epoch": 6.113064322812625,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 15301
    },
    {
      "epoch": 6.113463843387935,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 15302
    },
    {
      "epoch": 6.113863363963244,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 15303
    },
    {
      "epoch": 6.114262884538554,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 15304
    },
    {
      "epoch": 6.114662405113863,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 15305
    },
    {
      "epoch": 6.115061925689173,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 15306
    },
    {
      "epoch": 6.115461446264483,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 15307
    },
    {
      "epoch": 6.115860966839792,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 15308
    },
    {
      "epoch": 6.116260487415102,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 15309
    },
    {
      "epoch": 6.116660007990411,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 15310
    },
    {
      "epoch": 6.117059528565721,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 15311
    },
    {
      "epoch": 6.117459049141031,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 15312
    },
    {
      "epoch": 6.117858569716341,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 15313
    },
    {
      "epoch": 6.11825809029165,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 15314
    },
    {
      "epoch": 6.1186576108669595,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 15315
    },
    {
      "epoch": 6.119057131442269,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 15316
    },
    {
      "epoch": 6.119456652017579,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 15317
    },
    {
      "epoch": 6.119856172592889,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 15318
    },
    {
      "epoch": 6.120255693168198,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 15319
    },
    {
      "epoch": 6.1206552137435075,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 15320
    },
    {
      "epoch": 6.121054734318817,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 15321
    },
    {
      "epoch": 6.121454254894127,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 15322
    },
    {
      "epoch": 6.121853775469437,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 15323
    },
    {
      "epoch": 6.122253296044747,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 15324
    },
    {
      "epoch": 6.122652816620056,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 15325
    },
    {
      "epoch": 6.123052337195365,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 15326
    },
    {
      "epoch": 6.123451857770675,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 15327
    },
    {
      "epoch": 6.123851378345985,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 15328
    },
    {
      "epoch": 6.124250898921295,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 15329
    },
    {
      "epoch": 6.1246504194966045,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 15330
    },
    {
      "epoch": 6.125049940071913,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 15331
    },
    {
      "epoch": 6.125449460647223,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 15332
    },
    {
      "epoch": 6.125848981222533,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 15333
    },
    {
      "epoch": 6.126248501797843,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 15334
    },
    {
      "epoch": 6.1266480223731525,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 15335
    },
    {
      "epoch": 6.127047542948461,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 15336
    },
    {
      "epoch": 6.127447063523771,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 15337
    },
    {
      "epoch": 6.127846584099081,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 15338
    },
    {
      "epoch": 6.128246104674391,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 15339
    },
    {
      "epoch": 6.128645625249701,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 15340
    },
    {
      "epoch": 6.12904514582501,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 15341
    },
    {
      "epoch": 6.129444666400319,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 15342
    },
    {
      "epoch": 6.129844186975629,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 15343
    },
    {
      "epoch": 6.130243707550939,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 15344
    },
    {
      "epoch": 6.130643228126249,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 15345
    },
    {
      "epoch": 6.131042748701558,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 15346
    },
    {
      "epoch": 6.131442269276867,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 15347
    },
    {
      "epoch": 6.131841789852177,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 15348
    },
    {
      "epoch": 6.132241310427487,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 15349
    },
    {
      "epoch": 6.132640831002797,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 15350
    },
    {
      "epoch": 6.1330403515781065,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 15351
    },
    {
      "epoch": 6.133439872153416,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 15352
    },
    {
      "epoch": 6.133839392728725,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 15353
    },
    {
      "epoch": 6.134238913304035,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 15354
    },
    {
      "epoch": 6.134638433879345,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 15355
    },
    {
      "epoch": 6.1350379544546545,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 15356
    },
    {
      "epoch": 6.135437475029964,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 15357
    },
    {
      "epoch": 6.135836995605274,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 15358
    },
    {
      "epoch": 6.136236516180583,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 15359
    },
    {
      "epoch": 6.136636036755893,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 15360
    },
    {
      "epoch": 6.1370355573312025,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 15361
    },
    {
      "epoch": 6.137435077906512,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 15362
    },
    {
      "epoch": 6.137834598481822,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 15363
    },
    {
      "epoch": 6.138234119057131,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 15364
    },
    {
      "epoch": 6.138633639632441,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 15365
    },
    {
      "epoch": 6.139033160207751,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 15366
    },
    {
      "epoch": 6.13943268078306,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 15367
    },
    {
      "epoch": 6.13983220135837,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 15368
    },
    {
      "epoch": 6.14023172193368,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 15369
    },
    {
      "epoch": 6.140631242508989,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 15370
    },
    {
      "epoch": 6.141030763084299,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 15371
    },
    {
      "epoch": 6.141430283659608,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 15372
    },
    {
      "epoch": 6.141829804234918,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 15373
    },
    {
      "epoch": 6.142229324810228,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 15374
    },
    {
      "epoch": 6.142628845385538,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 15375
    },
    {
      "epoch": 6.143028365960847,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 15376
    },
    {
      "epoch": 6.1434278865361565,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 15377
    },
    {
      "epoch": 6.143827407111466,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 15378
    },
    {
      "epoch": 6.144226927686776,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 15379
    },
    {
      "epoch": 6.144626448262086,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 15380
    },
    {
      "epoch": 6.145025968837395,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 15381
    },
    {
      "epoch": 6.1454254894127045,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 15382
    },
    {
      "epoch": 6.145825009988014,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 15383
    },
    {
      "epoch": 6.146224530563324,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 15384
    },
    {
      "epoch": 6.146624051138634,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 15385
    },
    {
      "epoch": 6.147023571713944,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 15386
    },
    {
      "epoch": 6.1474230922892525,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 15387
    },
    {
      "epoch": 6.147822612864562,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 15388
    },
    {
      "epoch": 6.148222133439872,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 15389
    },
    {
      "epoch": 6.148621654015182,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 15390
    },
    {
      "epoch": 6.149021174590492,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 15391
    },
    {
      "epoch": 6.1494206951658015,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 15392
    },
    {
      "epoch": 6.14982021574111,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 15393
    },
    {
      "epoch": 6.15021973631642,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0595,
      "step": 15394
    },
    {
      "epoch": 6.15061925689173,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 15395
    },
    {
      "epoch": 6.15101877746704,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 15396
    },
    {
      "epoch": 6.1514182980423495,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 15397
    },
    {
      "epoch": 6.151817818617658,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 15398
    },
    {
      "epoch": 6.152217339192968,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 15399
    },
    {
      "epoch": 6.152616859768278,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 15400
    },
    {
      "epoch": 6.153016380343588,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 15401
    },
    {
      "epoch": 6.153415900918898,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 15402
    },
    {
      "epoch": 6.153815421494207,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 15403
    },
    {
      "epoch": 6.154214942069516,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 15404
    },
    {
      "epoch": 6.154614462644826,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 15405
    },
    {
      "epoch": 6.155013983220136,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 15406
    },
    {
      "epoch": 6.155413503795446,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 15407
    },
    {
      "epoch": 6.155813024370755,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 15408
    },
    {
      "epoch": 6.156212544946065,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 15409
    },
    {
      "epoch": 6.156612065521374,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 15410
    },
    {
      "epoch": 6.157011586096684,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 15411
    },
    {
      "epoch": 6.157411106671994,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 15412
    },
    {
      "epoch": 6.157810627247303,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 15413
    },
    {
      "epoch": 6.158210147822613,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 15414
    },
    {
      "epoch": 6.158609668397922,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 15415
    },
    {
      "epoch": 6.159009188973232,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 15416
    },
    {
      "epoch": 6.159408709548542,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 15417
    },
    {
      "epoch": 6.1598082301238515,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 15418
    },
    {
      "epoch": 6.160207750699161,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 15419
    },
    {
      "epoch": 6.160607271274471,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 15420
    },
    {
      "epoch": 6.16100679184978,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 15421
    },
    {
      "epoch": 6.16140631242509,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 15422
    },
    {
      "epoch": 6.1618058330003995,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 15423
    },
    {
      "epoch": 6.162205353575709,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 15424
    },
    {
      "epoch": 6.162604874151019,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 15425
    },
    {
      "epoch": 6.163004394726328,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 15426
    },
    {
      "epoch": 6.163403915301638,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 15427
    },
    {
      "epoch": 6.163803435876948,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 15428
    },
    {
      "epoch": 6.164202956452257,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 15429
    },
    {
      "epoch": 6.164602477027567,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 15430
    },
    {
      "epoch": 6.165001997602877,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 15431
    },
    {
      "epoch": 6.165401518178186,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 15432
    },
    {
      "epoch": 6.165801038753496,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 15433
    },
    {
      "epoch": 6.166200559328805,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 15434
    },
    {
      "epoch": 6.166600079904115,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 15435
    },
    {
      "epoch": 6.166999600479425,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0649,
      "step": 15436
    },
    {
      "epoch": 6.167399121054734,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 15437
    },
    {
      "epoch": 6.167798641630044,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 15438
    },
    {
      "epoch": 6.168198162205353,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 15439
    },
    {
      "epoch": 6.168597682780663,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 15440
    },
    {
      "epoch": 6.168997203355973,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 15441
    },
    {
      "epoch": 6.169396723931283,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 15442
    },
    {
      "epoch": 6.169796244506592,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 15443
    },
    {
      "epoch": 6.1701957650819015,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 15444
    },
    {
      "epoch": 6.170595285657211,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 15445
    },
    {
      "epoch": 6.170994806232521,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 15446
    },
    {
      "epoch": 6.171394326807831,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 15447
    },
    {
      "epoch": 6.171793847383141,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 15448
    },
    {
      "epoch": 6.1721933679584495,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 15449
    },
    {
      "epoch": 6.172592888533759,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 15450
    },
    {
      "epoch": 6.172992409109069,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 15451
    },
    {
      "epoch": 6.173391929684379,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 15452
    },
    {
      "epoch": 6.173791450259689,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 15453
    },
    {
      "epoch": 6.174190970834998,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 15454
    },
    {
      "epoch": 6.174590491410307,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 15455
    },
    {
      "epoch": 6.174990011985617,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 15456
    },
    {
      "epoch": 6.175389532560927,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 15457
    },
    {
      "epoch": 6.175789053136237,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 15458
    },
    {
      "epoch": 6.1761885737115465,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 15459
    },
    {
      "epoch": 6.176588094286855,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 15460
    },
    {
      "epoch": 6.176987614862165,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 15461
    },
    {
      "epoch": 6.177387135437475,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 15462
    },
    {
      "epoch": 6.177786656012785,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 15463
    },
    {
      "epoch": 6.1781861765880945,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 15464
    },
    {
      "epoch": 6.178585697163404,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 15465
    },
    {
      "epoch": 6.178985217738713,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 15466
    },
    {
      "epoch": 6.179384738314023,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 15467
    },
    {
      "epoch": 6.179784258889333,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 15468
    },
    {
      "epoch": 6.180183779464643,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 15469
    },
    {
      "epoch": 6.180583300039952,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 15470
    },
    {
      "epoch": 6.180982820615261,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 15471
    },
    {
      "epoch": 6.181382341190571,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 15472
    },
    {
      "epoch": 6.181781861765881,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 15473
    },
    {
      "epoch": 6.182181382341191,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 15474
    },
    {
      "epoch": 6.1825809029165,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 15475
    },
    {
      "epoch": 6.18298042349181,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 15476
    },
    {
      "epoch": 6.183379944067119,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 15477
    },
    {
      "epoch": 6.183779464642429,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 15478
    },
    {
      "epoch": 6.184178985217739,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 15479
    },
    {
      "epoch": 6.1845785057930485,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 15480
    },
    {
      "epoch": 6.184978026368358,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 15481
    },
    {
      "epoch": 6.185377546943668,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 15482
    },
    {
      "epoch": 6.185777067518977,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 15483
    },
    {
      "epoch": 6.186176588094287,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 15484
    },
    {
      "epoch": 6.1865761086695965,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 15485
    },
    {
      "epoch": 6.186975629244906,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 15486
    },
    {
      "epoch": 6.187375149820216,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 15487
    },
    {
      "epoch": 6.187774670395525,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 15488
    },
    {
      "epoch": 6.188174190970835,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 15489
    },
    {
      "epoch": 6.1885737115461446,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 15490
    },
    {
      "epoch": 6.188973232121454,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 15491
    },
    {
      "epoch": 6.189372752696764,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 15492
    },
    {
      "epoch": 6.189772273272074,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 15493
    },
    {
      "epoch": 6.190171793847383,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 15494
    },
    {
      "epoch": 6.190571314422693,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 15495
    },
    {
      "epoch": 6.190970834998002,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 15496
    },
    {
      "epoch": 6.191370355573312,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 15497
    },
    {
      "epoch": 6.191769876148622,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 15498
    },
    {
      "epoch": 6.192169396723932,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 15499
    },
    {
      "epoch": 6.192568917299241,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 15500
    },
    {
      "epoch": 6.19296843787455,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 15501
    },
    {
      "epoch": 6.19336795844986,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 15502
    },
    {
      "epoch": 6.19376747902517,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 15503
    },
    {
      "epoch": 6.19416699960048,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 15504
    },
    {
      "epoch": 6.194566520175789,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 15505
    },
    {
      "epoch": 6.1949660407510985,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 15506
    },
    {
      "epoch": 6.195365561326408,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 15507
    },
    {
      "epoch": 6.195765081901718,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 15508
    },
    {
      "epoch": 6.196164602477028,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 15509
    },
    {
      "epoch": 6.196564123052338,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 15510
    },
    {
      "epoch": 6.1969636436276465,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 15511
    },
    {
      "epoch": 6.197363164202956,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 15512
    },
    {
      "epoch": 6.197762684778266,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 15513
    },
    {
      "epoch": 6.198162205353576,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 15514
    },
    {
      "epoch": 6.198561725928886,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 15515
    },
    {
      "epoch": 6.1989612465041946,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 15516
    },
    {
      "epoch": 6.199360767079504,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 15517
    },
    {
      "epoch": 6.199760287654814,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 15518
    },
    {
      "epoch": 6.200159808230124,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 15519
    },
    {
      "epoch": 6.200559328805434,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 15520
    },
    {
      "epoch": 6.2009588493807435,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 15521
    },
    {
      "epoch": 6.201358369956052,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 15522
    },
    {
      "epoch": 6.201757890531362,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 15523
    },
    {
      "epoch": 6.202157411106672,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 15524
    },
    {
      "epoch": 6.202556931681982,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 15525
    },
    {
      "epoch": 6.2029564522572915,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 15526
    },
    {
      "epoch": 6.2033559728326,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 15527
    },
    {
      "epoch": 6.20375549340791,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 15528
    },
    {
      "epoch": 6.20415501398322,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 15529
    },
    {
      "epoch": 6.20455453455853,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 15530
    },
    {
      "epoch": 6.20495405513384,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 15531
    },
    {
      "epoch": 6.205353575709149,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 15532
    },
    {
      "epoch": 6.205753096284458,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 15533
    },
    {
      "epoch": 6.206152616859768,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 15534
    },
    {
      "epoch": 6.206552137435078,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 15535
    },
    {
      "epoch": 6.206951658010388,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 15536
    },
    {
      "epoch": 6.207351178585697,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 15537
    },
    {
      "epoch": 6.207750699161007,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 15538
    },
    {
      "epoch": 6.208150219736316,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 15539
    },
    {
      "epoch": 6.208549740311626,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 15540
    },
    {
      "epoch": 6.208949260886936,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 15541
    },
    {
      "epoch": 6.2093487814622454,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 15542
    },
    {
      "epoch": 6.209748302037555,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 15543
    },
    {
      "epoch": 6.210147822612864,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 15544
    },
    {
      "epoch": 6.210547343188174,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 15545
    },
    {
      "epoch": 6.210946863763484,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 15546
    },
    {
      "epoch": 6.2113463843387935,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 15547
    },
    {
      "epoch": 6.211745904914103,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 15548
    },
    {
      "epoch": 6.212145425489413,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 15549
    },
    {
      "epoch": 6.212544946064722,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 15550
    },
    {
      "epoch": 6.212944466640032,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 15551
    },
    {
      "epoch": 6.2133439872153415,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 15552
    },
    {
      "epoch": 6.213743507790651,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 15553
    },
    {
      "epoch": 6.214143028365961,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 15554
    },
    {
      "epoch": 6.214542548941271,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 15555
    },
    {
      "epoch": 6.21494206951658,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 15556
    },
    {
      "epoch": 6.21534159009189,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 15557
    },
    {
      "epoch": 6.215741110667199,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 15558
    },
    {
      "epoch": 6.216140631242509,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 15559
    },
    {
      "epoch": 6.216540151817819,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 15560
    },
    {
      "epoch": 6.216939672393128,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 15561
    },
    {
      "epoch": 6.217339192968438,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 15562
    },
    {
      "epoch": 6.217738713543747,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 15563
    },
    {
      "epoch": 6.218138234119057,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 15564
    },
    {
      "epoch": 6.218537754694367,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 15565
    },
    {
      "epoch": 6.218937275269677,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 15566
    },
    {
      "epoch": 6.219336795844986,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 15567
    },
    {
      "epoch": 6.2197363164202955,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 15568
    },
    {
      "epoch": 6.220135836995605,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 15569
    },
    {
      "epoch": 6.220535357570915,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 15570
    },
    {
      "epoch": 6.220934878146225,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 15571
    },
    {
      "epoch": 6.221334398721535,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 15572
    },
    {
      "epoch": 6.2217339192968435,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 15573
    },
    {
      "epoch": 6.222133439872153,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 15574
    },
    {
      "epoch": 6.222532960447463,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 15575
    },
    {
      "epoch": 6.222932481022773,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 15576
    },
    {
      "epoch": 6.223332001598083,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 15577
    },
    {
      "epoch": 6.2237315221733915,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 15578
    },
    {
      "epoch": 6.224131042748701,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 15579
    },
    {
      "epoch": 6.224530563324011,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 15580
    },
    {
      "epoch": 6.224930083899321,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 15581
    },
    {
      "epoch": 6.225329604474631,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 15582
    },
    {
      "epoch": 6.2257291250499405,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 15583
    },
    {
      "epoch": 6.226128645625249,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 15584
    },
    {
      "epoch": 6.226528166200559,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 15585
    },
    {
      "epoch": 6.226927686775869,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 15586
    },
    {
      "epoch": 6.227327207351179,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 15587
    },
    {
      "epoch": 6.2277267279264885,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 15588
    },
    {
      "epoch": 6.228126248501797,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 15589
    },
    {
      "epoch": 6.228525769077107,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 15590
    },
    {
      "epoch": 6.228925289652417,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 15591
    },
    {
      "epoch": 6.229324810227727,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 15592
    },
    {
      "epoch": 6.229724330803037,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 15593
    },
    {
      "epoch": 6.230123851378346,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0566,
      "step": 15594
    },
    {
      "epoch": 6.230523371953655,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 15595
    },
    {
      "epoch": 6.230922892528965,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 15596
    },
    {
      "epoch": 6.231322413104275,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 15597
    },
    {
      "epoch": 6.231721933679585,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 15598
    },
    {
      "epoch": 6.232121454254894,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 15599
    },
    {
      "epoch": 6.232520974830204,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 15600
    },
    {
      "epoch": 6.232920495405513,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 15601
    },
    {
      "epoch": 6.233320015980823,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 15602
    },
    {
      "epoch": 6.233719536556133,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 15603
    },
    {
      "epoch": 6.234119057131442,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 15604
    },
    {
      "epoch": 6.234518577706752,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 15605
    },
    {
      "epoch": 6.234918098282061,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 15606
    },
    {
      "epoch": 6.235317618857371,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 15607
    },
    {
      "epoch": 6.235717139432681,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 15608
    },
    {
      "epoch": 6.2361166600079905,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 15609
    },
    {
      "epoch": 6.2365161805833,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 15610
    },
    {
      "epoch": 6.23691570115861,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 15611
    },
    {
      "epoch": 6.237315221733919,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 15612
    },
    {
      "epoch": 6.237714742309229,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 15613
    },
    {
      "epoch": 6.2381142628845385,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 15614
    },
    {
      "epoch": 6.238513783459848,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 15615
    },
    {
      "epoch": 6.238913304035158,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 15616
    },
    {
      "epoch": 6.239312824610467,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 15617
    },
    {
      "epoch": 6.239712345185777,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 15618
    },
    {
      "epoch": 6.240111865761087,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 15619
    },
    {
      "epoch": 6.240511386336396,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 15620
    },
    {
      "epoch": 6.240910906911706,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 15621
    },
    {
      "epoch": 6.241310427487016,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 15622
    },
    {
      "epoch": 6.241709948062325,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 15623
    },
    {
      "epoch": 6.242109468637635,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 15624
    },
    {
      "epoch": 6.242508989212944,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 15625
    },
    {
      "epoch": 6.242908509788254,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 15626
    },
    {
      "epoch": 6.243308030363564,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 15627
    },
    {
      "epoch": 6.243707550938874,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 15628
    },
    {
      "epoch": 6.244107071514183,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 15629
    },
    {
      "epoch": 6.244506592089492,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 15630
    },
    {
      "epoch": 6.244906112664802,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 15631
    },
    {
      "epoch": 6.245305633240112,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 15632
    },
    {
      "epoch": 6.245705153815422,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 15633
    },
    {
      "epoch": 6.246104674390731,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 15634
    },
    {
      "epoch": 6.2465041949660405,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 15635
    },
    {
      "epoch": 6.24690371554135,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 15636
    },
    {
      "epoch": 6.24730323611666,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 15637
    },
    {
      "epoch": 6.24770275669197,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 15638
    },
    {
      "epoch": 6.24810227726728,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 15639
    },
    {
      "epoch": 6.2485017978425885,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 15640
    },
    {
      "epoch": 6.248901318417898,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 15641
    },
    {
      "epoch": 6.249300838993208,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 15642
    },
    {
      "epoch": 6.249700359568518,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 15643
    },
    {
      "epoch": 6.250099880143828,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 15644
    },
    {
      "epoch": 6.2504994007191375,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 15645
    },
    {
      "epoch": 6.250898921294446,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 15646
    },
    {
      "epoch": 6.251298441869756,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 15647
    },
    {
      "epoch": 6.251697962445066,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 15648
    },
    {
      "epoch": 6.252097483020376,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 15649
    },
    {
      "epoch": 6.2524970035956855,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 15650
    },
    {
      "epoch": 6.252896524170994,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 15651
    },
    {
      "epoch": 6.253296044746304,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 15652
    },
    {
      "epoch": 6.253695565321614,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 15653
    },
    {
      "epoch": 6.254095085896924,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 15654
    },
    {
      "epoch": 6.2544946064722335,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 15655
    },
    {
      "epoch": 6.254894127047543,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 15656
    },
    {
      "epoch": 6.255293647622852,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 15657
    },
    {
      "epoch": 6.255693168198162,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 15658
    },
    {
      "epoch": 6.256092688773472,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 15659
    },
    {
      "epoch": 6.256492209348782,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 15660
    },
    {
      "epoch": 6.256891729924091,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 15661
    },
    {
      "epoch": 6.257291250499401,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 15662
    },
    {
      "epoch": 6.25769077107471,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 15663
    },
    {
      "epoch": 6.25809029165002,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 15664
    },
    {
      "epoch": 6.25848981222533,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 15665
    },
    {
      "epoch": 6.258889332800639,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 15666
    },
    {
      "epoch": 6.259288853375949,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 15667
    },
    {
      "epoch": 6.259688373951258,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 15668
    },
    {
      "epoch": 6.260087894526568,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 15669
    },
    {
      "epoch": 6.260487415101878,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 15670
    },
    {
      "epoch": 6.2608869356771875,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 15671
    },
    {
      "epoch": 6.261286456252497,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 15672
    },
    {
      "epoch": 6.261685976827807,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 15673
    },
    {
      "epoch": 6.262085497403116,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 15674
    },
    {
      "epoch": 6.262485017978426,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 15675
    },
    {
      "epoch": 6.2628845385537355,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 15676
    },
    {
      "epoch": 6.263284059129045,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 15677
    },
    {
      "epoch": 6.263683579704355,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 15678
    },
    {
      "epoch": 6.264083100279665,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 15679
    },
    {
      "epoch": 6.264482620854974,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 15680
    },
    {
      "epoch": 6.2648821414302835,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 15681
    },
    {
      "epoch": 6.265281662005593,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 15682
    },
    {
      "epoch": 6.265681182580903,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 15683
    },
    {
      "epoch": 6.266080703156213,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 15684
    },
    {
      "epoch": 6.266480223731522,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 15685
    },
    {
      "epoch": 6.266879744306832,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 15686
    },
    {
      "epoch": 6.267279264882141,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 15687
    },
    {
      "epoch": 6.267678785457451,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 15688
    },
    {
      "epoch": 6.268078306032761,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 15689
    },
    {
      "epoch": 6.26847782660807,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 15690
    },
    {
      "epoch": 6.26887734718338,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 15691
    },
    {
      "epoch": 6.269276867758689,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 15692
    },
    {
      "epoch": 6.269676388333999,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 15693
    },
    {
      "epoch": 6.270075908909309,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 15694
    },
    {
      "epoch": 6.270475429484619,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 15695
    },
    {
      "epoch": 6.270874950059928,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 15696
    },
    {
      "epoch": 6.2712744706352375,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 15697
    },
    {
      "epoch": 6.271673991210547,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 15698
    },
    {
      "epoch": 6.272073511785857,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 15699
    },
    {
      "epoch": 6.272473032361167,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 15700
    },
    {
      "epoch": 6.272872552936477,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 15701
    },
    {
      "epoch": 6.2732720735117855,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 15702
    },
    {
      "epoch": 6.273671594087095,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 15703
    },
    {
      "epoch": 6.274071114662405,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 15704
    },
    {
      "epoch": 6.274470635237715,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 15705
    },
    {
      "epoch": 6.274870155813025,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 15706
    },
    {
      "epoch": 6.2752696763883336,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 15707
    },
    {
      "epoch": 6.275669196963643,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 15708
    },
    {
      "epoch": 6.276068717538953,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 15709
    },
    {
      "epoch": 6.276468238114263,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 15710
    },
    {
      "epoch": 6.276867758689573,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 15711
    },
    {
      "epoch": 6.2772672792648825,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 15712
    },
    {
      "epoch": 6.277666799840191,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 15713
    },
    {
      "epoch": 6.278066320415501,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 15714
    },
    {
      "epoch": 6.278465840990811,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 15715
    },
    {
      "epoch": 6.278865361566121,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 15716
    },
    {
      "epoch": 6.2792648821414305,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 15717
    },
    {
      "epoch": 6.27966440271674,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 15718
    },
    {
      "epoch": 6.280063923292049,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 15719
    },
    {
      "epoch": 6.280463443867359,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 15720
    },
    {
      "epoch": 6.280862964442669,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 15721
    },
    {
      "epoch": 6.281262485017979,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 15722
    },
    {
      "epoch": 6.281662005593288,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 15723
    },
    {
      "epoch": 6.282061526168597,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 15724
    },
    {
      "epoch": 6.282461046743907,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 15725
    },
    {
      "epoch": 6.282860567319217,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 15726
    },
    {
      "epoch": 6.283260087894527,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 15727
    },
    {
      "epoch": 6.283659608469836,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 15728
    },
    {
      "epoch": 6.284059129045146,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 15729
    },
    {
      "epoch": 6.284458649620455,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 15730
    },
    {
      "epoch": 6.284858170195765,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 15731
    },
    {
      "epoch": 6.285257690771075,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 15732
    },
    {
      "epoch": 6.285657211346384,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 15733
    },
    {
      "epoch": 6.286056731921694,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 15734
    },
    {
      "epoch": 6.286456252497004,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 15735
    },
    {
      "epoch": 6.286855773072313,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 15736
    },
    {
      "epoch": 6.287255293647623,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 15737
    },
    {
      "epoch": 6.2876548142229325,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 15738
    },
    {
      "epoch": 6.288054334798242,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 15739
    },
    {
      "epoch": 6.288453855373552,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 15740
    },
    {
      "epoch": 6.288853375948861,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 15741
    },
    {
      "epoch": 6.289252896524171,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 15742
    },
    {
      "epoch": 6.2896524170994805,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 15743
    },
    {
      "epoch": 6.29005193767479,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 15744
    },
    {
      "epoch": 6.2904514582501,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 15745
    },
    {
      "epoch": 6.29085097882541,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 15746
    },
    {
      "epoch": 6.291250499400719,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 15747
    },
    {
      "epoch": 6.291650019976029,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 15748
    },
    {
      "epoch": 6.292049540551338,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 15749
    },
    {
      "epoch": 6.292449061126648,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 15750
    },
    {
      "epoch": 6.292848581701958,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 15751
    },
    {
      "epoch": 6.293248102277268,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 15752
    },
    {
      "epoch": 6.293647622852577,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 15753
    },
    {
      "epoch": 6.294047143427886,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 15754
    },
    {
      "epoch": 6.294446664003196,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 15755
    },
    {
      "epoch": 6.294846184578506,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 15756
    },
    {
      "epoch": 6.295245705153816,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 15757
    },
    {
      "epoch": 6.295645225729125,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 15758
    },
    {
      "epoch": 6.2960447463044344,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 15759
    },
    {
      "epoch": 6.296444266879744,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 15760
    },
    {
      "epoch": 6.296843787455054,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 15761
    },
    {
      "epoch": 6.297243308030364,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 15762
    },
    {
      "epoch": 6.297642828605674,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 15763
    },
    {
      "epoch": 6.2980423491809825,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 15764
    },
    {
      "epoch": 6.298441869756292,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 15765
    },
    {
      "epoch": 6.298841390331602,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 15766
    },
    {
      "epoch": 6.299240910906912,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 15767
    },
    {
      "epoch": 6.299640431482222,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 15768
    },
    {
      "epoch": 6.300039952057531,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 15769
    },
    {
      "epoch": 6.30043947263284,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 15770
    },
    {
      "epoch": 6.30083899320815,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 15771
    },
    {
      "epoch": 6.30123851378346,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 15772
    },
    {
      "epoch": 6.30163803435877,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 15773
    },
    {
      "epoch": 6.3020375549340795,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 15774
    },
    {
      "epoch": 6.302437075509388,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 15775
    },
    {
      "epoch": 6.302836596084698,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 15776
    },
    {
      "epoch": 6.303236116660008,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 15777
    },
    {
      "epoch": 6.303635637235318,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 15778
    },
    {
      "epoch": 6.3040351578106275,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 15779
    },
    {
      "epoch": 6.304434678385936,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0584,
      "step": 15780
    },
    {
      "epoch": 6.304834198961246,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 15781
    },
    {
      "epoch": 6.305233719536556,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 15782
    },
    {
      "epoch": 6.305633240111866,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 15783
    },
    {
      "epoch": 6.3060327606871756,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 15784
    },
    {
      "epoch": 6.306432281262485,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 15785
    },
    {
      "epoch": 6.306831801837794,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 15786
    },
    {
      "epoch": 6.307231322413104,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 15787
    },
    {
      "epoch": 6.307630842988414,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 15788
    },
    {
      "epoch": 6.308030363563724,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 15789
    },
    {
      "epoch": 6.308429884139033,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 15790
    },
    {
      "epoch": 6.308829404714343,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 15791
    },
    {
      "epoch": 6.309228925289652,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 15792
    },
    {
      "epoch": 6.309628445864962,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 15793
    },
    {
      "epoch": 6.310027966440272,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 15794
    },
    {
      "epoch": 6.310427487015581,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 15795
    },
    {
      "epoch": 6.310827007590891,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 15796
    },
    {
      "epoch": 6.3112265281662,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 15797
    },
    {
      "epoch": 6.31162604874151,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 15798
    },
    {
      "epoch": 6.31202556931682,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 15799
    },
    {
      "epoch": 6.3124250898921295,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 15800
    },
    {
      "epoch": 6.312824610467439,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 15801
    },
    {
      "epoch": 6.313224131042749,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 15802
    },
    {
      "epoch": 6.313623651618058,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 15803
    },
    {
      "epoch": 6.314023172193368,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 15804
    },
    {
      "epoch": 6.3144226927686775,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 15805
    },
    {
      "epoch": 6.314822213343987,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 15806
    },
    {
      "epoch": 6.315221733919297,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 15807
    },
    {
      "epoch": 6.315621254494607,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 15808
    },
    {
      "epoch": 6.316020775069916,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 15809
    },
    {
      "epoch": 6.316420295645226,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 15810
    },
    {
      "epoch": 6.316819816220535,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 15811
    },
    {
      "epoch": 6.317219336795845,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 15812
    },
    {
      "epoch": 6.317618857371155,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 15813
    },
    {
      "epoch": 6.318018377946464,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0559,
      "step": 15814
    },
    {
      "epoch": 6.318417898521774,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 15815
    },
    {
      "epoch": 6.318817419097083,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 15816
    },
    {
      "epoch": 6.319216939672393,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 15817
    },
    {
      "epoch": 6.319616460247703,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 15818
    },
    {
      "epoch": 6.320015980823013,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 15819
    },
    {
      "epoch": 6.320415501398322,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 15820
    },
    {
      "epoch": 6.320815021973631,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 15821
    },
    {
      "epoch": 6.321214542548941,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 15822
    },
    {
      "epoch": 6.321614063124251,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 15823
    },
    {
      "epoch": 6.322013583699561,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 15824
    },
    {
      "epoch": 6.322413104274871,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0588,
      "step": 15825
    },
    {
      "epoch": 6.3228126248501795,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 15826
    },
    {
      "epoch": 6.323212145425489,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 15827
    },
    {
      "epoch": 6.323611666000799,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 15828
    },
    {
      "epoch": 6.324011186576109,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 15829
    },
    {
      "epoch": 6.324410707151419,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 15830
    },
    {
      "epoch": 6.3248102277267275,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 15831
    },
    {
      "epoch": 6.325209748302037,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 15832
    },
    {
      "epoch": 6.325609268877347,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 15833
    },
    {
      "epoch": 6.326008789452657,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 15834
    },
    {
      "epoch": 6.326408310027967,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 15835
    },
    {
      "epoch": 6.3268078306032765,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 15836
    },
    {
      "epoch": 6.327207351178585,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 15837
    },
    {
      "epoch": 6.327606871753895,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 15838
    },
    {
      "epoch": 6.328006392329205,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 15839
    },
    {
      "epoch": 6.328405912904515,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 15840
    },
    {
      "epoch": 6.3288054334798245,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 15841
    },
    {
      "epoch": 6.329204954055134,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 15842
    },
    {
      "epoch": 6.329604474630443,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 15843
    },
    {
      "epoch": 6.330003995205753,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 15844
    },
    {
      "epoch": 6.330403515781063,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 15845
    },
    {
      "epoch": 6.3308030363563725,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 15846
    },
    {
      "epoch": 6.331202556931682,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 15847
    },
    {
      "epoch": 6.331602077506991,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 15848
    },
    {
      "epoch": 6.332001598082301,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 15849
    },
    {
      "epoch": 6.332401118657611,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 15850
    },
    {
      "epoch": 6.332800639232921,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 15851
    },
    {
      "epoch": 6.33320015980823,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 15852
    },
    {
      "epoch": 6.33359968038354,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 15853
    },
    {
      "epoch": 6.333999200958849,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 15854
    },
    {
      "epoch": 6.334398721534159,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 15855
    },
    {
      "epoch": 6.334798242109469,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 15856
    },
    {
      "epoch": 6.335197762684778,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 15857
    },
    {
      "epoch": 6.335597283260088,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 15858
    },
    {
      "epoch": 6.335996803835398,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 15859
    },
    {
      "epoch": 6.336396324410707,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 15860
    },
    {
      "epoch": 6.336795844986017,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 15861
    },
    {
      "epoch": 6.3371953655613265,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 15862
    },
    {
      "epoch": 6.337594886136636,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 15863
    },
    {
      "epoch": 6.337994406711946,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 15864
    },
    {
      "epoch": 6.338393927287255,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 15865
    },
    {
      "epoch": 6.338793447862565,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 15866
    },
    {
      "epoch": 6.3391929684378745,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 15867
    },
    {
      "epoch": 6.339592489013184,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 15868
    },
    {
      "epoch": 6.339992009588494,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 15869
    },
    {
      "epoch": 6.340391530163803,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 15870
    },
    {
      "epoch": 6.340791050739113,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 15871
    },
    {
      "epoch": 6.3411905713144225,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 15872
    },
    {
      "epoch": 6.341590091889732,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 15873
    },
    {
      "epoch": 6.341989612465042,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 15874
    },
    {
      "epoch": 6.342389133040352,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 15875
    },
    {
      "epoch": 6.342788653615661,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 15876
    },
    {
      "epoch": 6.343188174190971,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 15877
    },
    {
      "epoch": 6.34358769476628,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 15878
    },
    {
      "epoch": 6.34398721534159,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 15879
    },
    {
      "epoch": 6.3443867359169,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 15880
    },
    {
      "epoch": 6.34478625649221,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 15881
    },
    {
      "epoch": 6.345185777067519,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 15882
    },
    {
      "epoch": 6.345585297642828,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 15883
    },
    {
      "epoch": 6.345984818218138,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 15884
    },
    {
      "epoch": 6.346384338793448,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 15885
    },
    {
      "epoch": 6.346783859368758,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 15886
    },
    {
      "epoch": 6.347183379944067,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 15887
    },
    {
      "epoch": 6.3475829005193765,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 15888
    },
    {
      "epoch": 6.347982421094686,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 15889
    },
    {
      "epoch": 6.348381941669996,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 15890
    },
    {
      "epoch": 6.348781462245306,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 15891
    },
    {
      "epoch": 6.349180982820616,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 15892
    },
    {
      "epoch": 6.3495805033959245,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 15893
    },
    {
      "epoch": 6.349980023971234,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 15894
    },
    {
      "epoch": 6.350379544546544,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 15895
    },
    {
      "epoch": 6.350779065121854,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 15896
    },
    {
      "epoch": 6.351178585697164,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 15897
    },
    {
      "epoch": 6.351578106272473,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 15898
    },
    {
      "epoch": 6.351977626847782,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 15899
    },
    {
      "epoch": 6.352377147423092,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 15900
    },
    {
      "epoch": 6.352776667998402,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 15901
    },
    {
      "epoch": 6.353176188573712,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 15902
    },
    {
      "epoch": 6.3535757091490215,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 15903
    },
    {
      "epoch": 6.35397522972433,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 15904
    },
    {
      "epoch": 6.35437475029964,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 15905
    },
    {
      "epoch": 6.35477427087495,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 15906
    },
    {
      "epoch": 6.35517379145026,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 15907
    },
    {
      "epoch": 6.3555733120255695,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 15908
    },
    {
      "epoch": 6.355972832600879,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 15909
    },
    {
      "epoch": 6.356372353176188,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 15910
    },
    {
      "epoch": 6.356771873751498,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 15911
    },
    {
      "epoch": 6.357171394326808,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 15912
    },
    {
      "epoch": 6.357570914902118,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 15913
    },
    {
      "epoch": 6.357970435477427,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 15914
    },
    {
      "epoch": 6.358369956052737,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 15915
    },
    {
      "epoch": 6.358769476628046,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 15916
    },
    {
      "epoch": 6.359168997203356,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 15917
    },
    {
      "epoch": 6.359568517778666,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 15918
    },
    {
      "epoch": 6.359968038353975,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 15919
    },
    {
      "epoch": 6.360367558929285,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 15920
    },
    {
      "epoch": 6.360767079504594,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 15921
    },
    {
      "epoch": 6.361166600079904,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 15922
    },
    {
      "epoch": 6.361566120655214,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 15923
    },
    {
      "epoch": 6.361965641230523,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 15924
    },
    {
      "epoch": 6.362365161805833,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 15925
    },
    {
      "epoch": 6.362764682381143,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 15926
    },
    {
      "epoch": 6.363164202956452,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 15927
    },
    {
      "epoch": 6.363563723531762,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 15928
    },
    {
      "epoch": 6.3639632441070715,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 15929
    },
    {
      "epoch": 6.364362764682381,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 15930
    },
    {
      "epoch": 6.364762285257691,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 15931
    },
    {
      "epoch": 6.365161805833001,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 15932
    },
    {
      "epoch": 6.36556132640831,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 15933
    },
    {
      "epoch": 6.3659608469836195,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 15934
    },
    {
      "epoch": 6.366360367558929,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 15935
    },
    {
      "epoch": 6.366759888134239,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 15936
    },
    {
      "epoch": 6.367159408709549,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 15937
    },
    {
      "epoch": 6.367558929284858,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 15938
    },
    {
      "epoch": 6.367958449860168,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 15939
    },
    {
      "epoch": 6.368357970435477,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 15940
    },
    {
      "epoch": 6.368757491010787,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 15941
    },
    {
      "epoch": 6.369157011586097,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 15942
    },
    {
      "epoch": 6.369556532161407,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 15943
    },
    {
      "epoch": 6.369956052736716,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 15944
    },
    {
      "epoch": 6.370355573312025,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 15945
    },
    {
      "epoch": 6.370755093887335,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 15946
    },
    {
      "epoch": 6.371154614462645,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 15947
    },
    {
      "epoch": 6.371554135037955,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 15948
    },
    {
      "epoch": 6.3719536556132645,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 15949
    },
    {
      "epoch": 6.372353176188573,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 15950
    },
    {
      "epoch": 6.372752696763883,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 15951
    },
    {
      "epoch": 6.373152217339193,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 15952
    },
    {
      "epoch": 6.373551737914503,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 15953
    },
    {
      "epoch": 6.373951258489813,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 15954
    },
    {
      "epoch": 6.3743507790651215,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 15955
    },
    {
      "epoch": 6.374750299640431,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 15956
    },
    {
      "epoch": 6.375149820215741,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 15957
    },
    {
      "epoch": 6.375549340791051,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 15958
    },
    {
      "epoch": 6.375948861366361,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 15959
    },
    {
      "epoch": 6.3763483819416695,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 15960
    },
    {
      "epoch": 6.376747902516979,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 15961
    },
    {
      "epoch": 6.377147423092289,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 15962
    },
    {
      "epoch": 6.377546943667599,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 15963
    },
    {
      "epoch": 6.377946464242909,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 15964
    },
    {
      "epoch": 6.3783459848182185,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 15965
    },
    {
      "epoch": 6.378745505393527,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 15966
    },
    {
      "epoch": 6.379145025968837,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 15967
    },
    {
      "epoch": 6.379544546544147,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 15968
    },
    {
      "epoch": 6.379944067119457,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 15969
    },
    {
      "epoch": 6.3803435876947665,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 15970
    },
    {
      "epoch": 6.380743108270076,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 15971
    },
    {
      "epoch": 6.381142628845385,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 15972
    },
    {
      "epoch": 6.381542149420695,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 15973
    },
    {
      "epoch": 6.381941669996005,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 15974
    },
    {
      "epoch": 6.3823411905713145,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 15975
    },
    {
      "epoch": 6.382740711146624,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 15976
    },
    {
      "epoch": 6.383140231721933,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 15977
    },
    {
      "epoch": 6.383539752297243,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 15978
    },
    {
      "epoch": 6.383939272872553,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 15979
    },
    {
      "epoch": 6.384338793447863,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 15980
    },
    {
      "epoch": 6.384738314023172,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 15981
    },
    {
      "epoch": 6.385137834598482,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 15982
    },
    {
      "epoch": 6.385537355173791,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 15983
    },
    {
      "epoch": 6.385936875749101,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 15984
    },
    {
      "epoch": 6.386336396324411,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 15985
    },
    {
      "epoch": 6.38673591689972,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 15986
    },
    {
      "epoch": 6.38713543747503,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 15987
    },
    {
      "epoch": 6.38753495805034,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 15988
    },
    {
      "epoch": 6.387934478625649,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 15989
    },
    {
      "epoch": 6.388333999200959,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 15990
    },
    {
      "epoch": 6.3887335197762685,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 15991
    },
    {
      "epoch": 6.389133040351578,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 15992
    },
    {
      "epoch": 6.389532560926888,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 15993
    },
    {
      "epoch": 6.389932081502197,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 15994
    },
    {
      "epoch": 6.390331602077507,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 15995
    },
    {
      "epoch": 6.3907311226528165,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 15996
    },
    {
      "epoch": 6.391130643228126,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 15997
    },
    {
      "epoch": 6.391530163803436,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 15998
    },
    {
      "epoch": 6.391929684378746,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 15999
    },
    {
      "epoch": 6.392329204954055,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 16000
    },
    {
      "epoch": 6.3927287255293646,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 16001
    },
    {
      "epoch": 6.393128246104674,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 16002
    },
    {
      "epoch": 6.393527766679984,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 16003
    },
    {
      "epoch": 6.393927287255294,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 16004
    },
    {
      "epoch": 6.394326807830604,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 16005
    },
    {
      "epoch": 6.394726328405913,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 16006
    },
    {
      "epoch": 6.395125848981222,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 16007
    },
    {
      "epoch": 6.395525369556532,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 16008
    },
    {
      "epoch": 6.395924890131842,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 16009
    },
    {
      "epoch": 6.396324410707152,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 16010
    },
    {
      "epoch": 6.396723931282461,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 16011
    },
    {
      "epoch": 6.39712345185777,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 16012
    },
    {
      "epoch": 6.39752297243308,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 16013
    },
    {
      "epoch": 6.39792249300839,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 16014
    },
    {
      "epoch": 6.3983220135837,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 16015
    },
    {
      "epoch": 6.39872153415901,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 16016
    },
    {
      "epoch": 6.3991210547343185,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 16017
    },
    {
      "epoch": 6.399520575309628,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 16018
    },
    {
      "epoch": 6.399920095884938,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 16019
    },
    {
      "epoch": 6.400319616460248,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 16020
    },
    {
      "epoch": 6.400719137035558,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 16021
    },
    {
      "epoch": 6.401118657610867,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 16022
    },
    {
      "epoch": 6.401518178186176,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 16023
    },
    {
      "epoch": 6.401917698761486,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 16024
    },
    {
      "epoch": 6.402317219336796,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 16025
    },
    {
      "epoch": 6.402716739912106,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 16026
    },
    {
      "epoch": 6.4031162604874154,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 16027
    },
    {
      "epoch": 6.403515781062724,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 16028
    },
    {
      "epoch": 6.403915301638034,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 16029
    },
    {
      "epoch": 6.404314822213344,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 16030
    },
    {
      "epoch": 6.404714342788654,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 16031
    },
    {
      "epoch": 6.4051138633639635,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 16032
    },
    {
      "epoch": 6.405513383939273,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 16033
    },
    {
      "epoch": 6.405912904514582,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 16034
    },
    {
      "epoch": 6.406312425089892,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 16035
    },
    {
      "epoch": 6.406711945665202,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 16036
    },
    {
      "epoch": 6.4071114662405115,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 16037
    },
    {
      "epoch": 6.407510986815821,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 16038
    },
    {
      "epoch": 6.407910507391131,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 16039
    },
    {
      "epoch": 6.40831002796644,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 16040
    },
    {
      "epoch": 6.40870954854175,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 16041
    },
    {
      "epoch": 6.40910906911706,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 16042
    },
    {
      "epoch": 6.409508589692369,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 16043
    },
    {
      "epoch": 6.409908110267679,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 16044
    },
    {
      "epoch": 6.410307630842988,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 16045
    },
    {
      "epoch": 6.410707151418298,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 16046
    },
    {
      "epoch": 6.411106671993608,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 16047
    },
    {
      "epoch": 6.411506192568917,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0631,
      "step": 16048
    },
    {
      "epoch": 6.411905713144227,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 16049
    },
    {
      "epoch": 6.412305233719536,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 16050
    },
    {
      "epoch": 6.412704754294846,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 16051
    },
    {
      "epoch": 6.413104274870156,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 16052
    },
    {
      "epoch": 6.4135037954454654,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0575,
      "step": 16053
    },
    {
      "epoch": 6.413903316020775,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 16054
    },
    {
      "epoch": 6.414302836596085,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 16055
    },
    {
      "epoch": 6.414702357171394,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 16056
    },
    {
      "epoch": 6.415101877746704,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 16057
    },
    {
      "epoch": 6.4155013983220135,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 16058
    },
    {
      "epoch": 6.415900918897323,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 16059
    },
    {
      "epoch": 6.416300439472633,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 16060
    },
    {
      "epoch": 6.416699960047943,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 16061
    },
    {
      "epoch": 6.417099480623252,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 16062
    },
    {
      "epoch": 6.4174990011985615,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 16063
    },
    {
      "epoch": 6.417898521773871,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 16064
    },
    {
      "epoch": 6.418298042349181,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 16065
    },
    {
      "epoch": 6.418697562924491,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 16066
    },
    {
      "epoch": 6.4190970834998,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 16067
    },
    {
      "epoch": 6.41949660407511,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 16068
    },
    {
      "epoch": 6.419896124650419,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 16069
    },
    {
      "epoch": 6.420295645225729,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 16070
    },
    {
      "epoch": 6.420695165801039,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 16071
    },
    {
      "epoch": 6.421094686376349,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 16072
    },
    {
      "epoch": 6.421494206951658,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 16073
    },
    {
      "epoch": 6.421893727526967,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 16074
    },
    {
      "epoch": 6.422293248102277,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 16075
    },
    {
      "epoch": 6.422692768677587,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 16076
    },
    {
      "epoch": 6.423092289252897,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 16077
    },
    {
      "epoch": 6.4234918098282066,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 16078
    },
    {
      "epoch": 6.4238913304035155,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 16079
    },
    {
      "epoch": 6.424290850978825,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 16080
    },
    {
      "epoch": 6.424690371554135,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 16081
    },
    {
      "epoch": 6.425089892129445,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 16082
    },
    {
      "epoch": 6.425489412704755,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 16083
    },
    {
      "epoch": 6.4258889332800635,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 16084
    },
    {
      "epoch": 6.426288453855373,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 16085
    },
    {
      "epoch": 6.426687974430683,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 16086
    },
    {
      "epoch": 6.427087495005993,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 16087
    },
    {
      "epoch": 6.427487015581303,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 16088
    },
    {
      "epoch": 6.427886536156612,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 16089
    },
    {
      "epoch": 6.428286056731921,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 16090
    },
    {
      "epoch": 6.428685577307231,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 16091
    },
    {
      "epoch": 6.429085097882541,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 16092
    },
    {
      "epoch": 6.429484618457851,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 16093
    },
    {
      "epoch": 6.4298841390331605,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 16094
    },
    {
      "epoch": 6.43028365960847,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 16095
    },
    {
      "epoch": 6.430683180183779,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 16096
    },
    {
      "epoch": 6.431082700759089,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 16097
    },
    {
      "epoch": 6.431482221334399,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 16098
    },
    {
      "epoch": 6.4318817419097085,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 16099
    },
    {
      "epoch": 6.432281262485018,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 16100
    },
    {
      "epoch": 6.432680783060327,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 16101
    },
    {
      "epoch": 6.433080303635637,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 16102
    },
    {
      "epoch": 6.433479824210947,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 16103
    },
    {
      "epoch": 6.433879344786257,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 16104
    },
    {
      "epoch": 6.434278865361566,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 16105
    },
    {
      "epoch": 6.434678385936876,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 16106
    },
    {
      "epoch": 6.435077906512185,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 16107
    },
    {
      "epoch": 6.435477427087495,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 16108
    },
    {
      "epoch": 6.435876947662805,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 16109
    },
    {
      "epoch": 6.436276468238114,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 16110
    },
    {
      "epoch": 6.436675988813424,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 16111
    },
    {
      "epoch": 6.437075509388734,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 16112
    },
    {
      "epoch": 6.437475029964043,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 16113
    },
    {
      "epoch": 6.437874550539353,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 16114
    },
    {
      "epoch": 6.438274071114662,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 16115
    },
    {
      "epoch": 6.438673591689972,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 16116
    },
    {
      "epoch": 6.439073112265282,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 16117
    },
    {
      "epoch": 6.439472632840591,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 16118
    },
    {
      "epoch": 6.439872153415901,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 16119
    },
    {
      "epoch": 6.4402716739912105,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 16120
    },
    {
      "epoch": 6.44067119456652,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 16121
    },
    {
      "epoch": 6.44107071514183,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 16122
    },
    {
      "epoch": 6.44147023571714,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 16123
    },
    {
      "epoch": 6.441869756292449,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 16124
    },
    {
      "epoch": 6.4422692768677585,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 16125
    },
    {
      "epoch": 6.442668797443068,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 16126
    },
    {
      "epoch": 6.443068318018378,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 16127
    },
    {
      "epoch": 6.443467838593688,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 16128
    },
    {
      "epoch": 6.443867359168998,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 16129
    },
    {
      "epoch": 6.444266879744307,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 16130
    },
    {
      "epoch": 6.444666400319616,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 16131
    },
    {
      "epoch": 6.445065920894926,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 16132
    },
    {
      "epoch": 6.445465441470236,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 16133
    },
    {
      "epoch": 6.445864962045546,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 16134
    },
    {
      "epoch": 6.446264482620855,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 16135
    },
    {
      "epoch": 6.446664003196164,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 16136
    },
    {
      "epoch": 6.447063523771474,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 16137
    },
    {
      "epoch": 6.447463044346784,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 16138
    },
    {
      "epoch": 6.447862564922094,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 16139
    },
    {
      "epoch": 6.448262085497403,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 16140
    },
    {
      "epoch": 6.448661606072712,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 16141
    },
    {
      "epoch": 6.449061126648022,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 16142
    },
    {
      "epoch": 6.449460647223332,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 16143
    },
    {
      "epoch": 6.449860167798642,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 16144
    },
    {
      "epoch": 6.450259688373952,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 16145
    },
    {
      "epoch": 6.4506592089492605,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 16146
    },
    {
      "epoch": 6.45105872952457,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 16147
    },
    {
      "epoch": 6.45145825009988,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 16148
    },
    {
      "epoch": 6.45185777067519,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 16149
    },
    {
      "epoch": 6.4522572912505,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 16150
    },
    {
      "epoch": 6.452656811825809,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 16151
    },
    {
      "epoch": 6.453056332401118,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 16152
    },
    {
      "epoch": 6.453455852976428,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 16153
    },
    {
      "epoch": 6.453855373551738,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 16154
    },
    {
      "epoch": 6.454254894127048,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 16155
    },
    {
      "epoch": 6.4546544147023575,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 16156
    },
    {
      "epoch": 6.455053935277666,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 16157
    },
    {
      "epoch": 6.455453455852976,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 16158
    },
    {
      "epoch": 6.455852976428286,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 16159
    },
    {
      "epoch": 6.456252497003596,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 16160
    },
    {
      "epoch": 6.4566520175789055,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 16161
    },
    {
      "epoch": 6.457051538154215,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 16162
    },
    {
      "epoch": 6.457451058729524,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 16163
    },
    {
      "epoch": 6.457850579304834,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 16164
    },
    {
      "epoch": 6.458250099880144,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 16165
    },
    {
      "epoch": 6.4586496204554535,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 16166
    },
    {
      "epoch": 6.459049141030763,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 16167
    },
    {
      "epoch": 6.459448661606073,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 16168
    },
    {
      "epoch": 6.459848182181382,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 16169
    },
    {
      "epoch": 6.460247702756692,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 16170
    },
    {
      "epoch": 6.460647223332002,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 16171
    },
    {
      "epoch": 6.461046743907311,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 16172
    },
    {
      "epoch": 6.461446264482621,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 16173
    },
    {
      "epoch": 6.46184578505793,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 16174
    },
    {
      "epoch": 6.46224530563324,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 16175
    },
    {
      "epoch": 6.46264482620855,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 16176
    },
    {
      "epoch": 6.463044346783859,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 16177
    },
    {
      "epoch": 6.463443867359169,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 16178
    },
    {
      "epoch": 6.463843387934479,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 16179
    },
    {
      "epoch": 6.464242908509788,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 16180
    },
    {
      "epoch": 6.464642429085098,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 16181
    },
    {
      "epoch": 6.4650419496604075,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 16182
    },
    {
      "epoch": 6.465441470235717,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 16183
    },
    {
      "epoch": 6.465840990811027,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 16184
    },
    {
      "epoch": 6.466240511386337,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 16185
    },
    {
      "epoch": 6.466640031961646,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 16186
    },
    {
      "epoch": 6.4670395525369555,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 16187
    },
    {
      "epoch": 6.467439073112265,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 16188
    },
    {
      "epoch": 6.467838593687575,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 16189
    },
    {
      "epoch": 6.468238114262885,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 16190
    },
    {
      "epoch": 6.468637634838194,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 16191
    },
    {
      "epoch": 6.4690371554135035,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 16192
    },
    {
      "epoch": 6.469436675988813,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 16193
    },
    {
      "epoch": 6.469836196564123,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 16194
    },
    {
      "epoch": 6.470235717139433,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 16195
    },
    {
      "epoch": 6.470635237714743,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 16196
    },
    {
      "epoch": 6.471034758290052,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 16197
    },
    {
      "epoch": 6.471434278865361,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 16198
    },
    {
      "epoch": 6.471833799440671,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 16199
    },
    {
      "epoch": 6.472233320015981,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 16200
    },
    {
      "epoch": 6.472632840591291,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 16201
    },
    {
      "epoch": 6.4730323611666005,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 16202
    },
    {
      "epoch": 6.473431881741909,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 16203
    },
    {
      "epoch": 6.473831402317219,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 16204
    },
    {
      "epoch": 6.474230922892529,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 16205
    },
    {
      "epoch": 6.474630443467839,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 16206
    },
    {
      "epoch": 6.475029964043149,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 16207
    },
    {
      "epoch": 6.4754294846184575,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 16208
    },
    {
      "epoch": 6.475829005193767,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 16209
    },
    {
      "epoch": 6.476228525769077,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 16210
    },
    {
      "epoch": 6.476628046344387,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 16211
    },
    {
      "epoch": 6.477027566919697,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 16212
    },
    {
      "epoch": 6.477427087495006,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 16213
    },
    {
      "epoch": 6.477826608070315,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 16214
    },
    {
      "epoch": 6.478226128645625,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 16215
    },
    {
      "epoch": 6.478625649220935,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 16216
    },
    {
      "epoch": 6.479025169796245,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 16217
    },
    {
      "epoch": 6.479424690371554,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 16218
    },
    {
      "epoch": 6.479824210946864,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 16219
    },
    {
      "epoch": 6.480223731522173,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 16220
    },
    {
      "epoch": 6.480623252097483,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 16221
    },
    {
      "epoch": 6.481022772672793,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 16222
    },
    {
      "epoch": 6.4814222932481025,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 16223
    },
    {
      "epoch": 6.481821813823412,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 16224
    },
    {
      "epoch": 6.482221334398721,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 16225
    },
    {
      "epoch": 6.482620854974031,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 16226
    },
    {
      "epoch": 6.483020375549341,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 16227
    },
    {
      "epoch": 6.4834198961246505,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 16228
    },
    {
      "epoch": 6.48381941669996,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 16229
    },
    {
      "epoch": 6.484218937275269,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 16230
    },
    {
      "epoch": 6.484618457850579,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 16231
    },
    {
      "epoch": 6.485017978425889,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 16232
    },
    {
      "epoch": 6.485417499001199,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 16233
    },
    {
      "epoch": 6.485817019576508,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 16234
    },
    {
      "epoch": 6.486216540151818,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 16235
    },
    {
      "epoch": 6.486616060727127,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 16236
    },
    {
      "epoch": 6.487015581302437,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 16237
    },
    {
      "epoch": 6.487415101877747,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 16238
    },
    {
      "epoch": 6.487814622453056,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 16239
    },
    {
      "epoch": 6.488214143028366,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 16240
    },
    {
      "epoch": 6.488613663603676,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 16241
    },
    {
      "epoch": 6.489013184178985,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 16242
    },
    {
      "epoch": 6.489412704754295,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 16243
    },
    {
      "epoch": 6.4898122253296044,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 16244
    },
    {
      "epoch": 6.490211745904914,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 16245
    },
    {
      "epoch": 6.490611266480224,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 16246
    },
    {
      "epoch": 6.491010787055533,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 16247
    },
    {
      "epoch": 6.491410307630843,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 16248
    },
    {
      "epoch": 6.4918098282061525,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 16249
    },
    {
      "epoch": 6.492209348781462,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 16250
    },
    {
      "epoch": 6.492608869356772,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0554,
      "step": 16251
    },
    {
      "epoch": 6.493008389932082,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 16252
    },
    {
      "epoch": 6.493407910507391,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 16253
    },
    {
      "epoch": 6.4938074310827005,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 16254
    },
    {
      "epoch": 6.49420695165801,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 16255
    },
    {
      "epoch": 6.49460647223332,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 16256
    },
    {
      "epoch": 6.49500599280863,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 16257
    },
    {
      "epoch": 6.49540551338394,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 16258
    },
    {
      "epoch": 6.495805033959249,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 16259
    },
    {
      "epoch": 6.496204554534558,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 16260
    },
    {
      "epoch": 6.496604075109868,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 16261
    },
    {
      "epoch": 6.497003595685178,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 16262
    },
    {
      "epoch": 6.497403116260488,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 16263
    },
    {
      "epoch": 6.497802636835797,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 16264
    },
    {
      "epoch": 6.498202157411106,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 16265
    },
    {
      "epoch": 6.498601677986416,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 16266
    },
    {
      "epoch": 6.499001198561726,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 16267
    },
    {
      "epoch": 6.499400719137036,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 16268
    },
    {
      "epoch": 6.4998002397123456,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 16269
    },
    {
      "epoch": 6.5001997602876544,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0603,
      "step": 16270
    },
    {
      "epoch": 6.500599280862964,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 16271
    },
    {
      "epoch": 6.500998801438274,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 16272
    },
    {
      "epoch": 6.501398322013584,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 16273
    },
    {
      "epoch": 6.501797842588894,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 16274
    },
    {
      "epoch": 6.502197363164203,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 16275
    },
    {
      "epoch": 6.502596883739512,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 16276
    },
    {
      "epoch": 6.502996404314822,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 16277
    },
    {
      "epoch": 6.503395924890132,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 16278
    },
    {
      "epoch": 6.503795445465442,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 16279
    },
    {
      "epoch": 6.504194966040751,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 16280
    },
    {
      "epoch": 6.50459448661606,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 16281
    },
    {
      "epoch": 6.50499400719137,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 16282
    },
    {
      "epoch": 6.50539352776668,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 16283
    },
    {
      "epoch": 6.50579304834199,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 16284
    },
    {
      "epoch": 6.5061925689172995,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 16285
    },
    {
      "epoch": 6.506592089492609,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 16286
    },
    {
      "epoch": 6.506991610067918,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 16287
    },
    {
      "epoch": 6.507391130643228,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 16288
    },
    {
      "epoch": 6.507790651218538,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 16289
    },
    {
      "epoch": 6.5081901717938475,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 16290
    },
    {
      "epoch": 6.508589692369157,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 16291
    },
    {
      "epoch": 6.508989212944467,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 16292
    },
    {
      "epoch": 6.509388733519776,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 16293
    },
    {
      "epoch": 6.509788254095086,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 16294
    },
    {
      "epoch": 6.5101877746703956,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 16295
    },
    {
      "epoch": 6.510587295245705,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 16296
    },
    {
      "epoch": 6.510986815821015,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 16297
    },
    {
      "epoch": 6.511386336396324,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 16298
    },
    {
      "epoch": 6.511785856971634,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 16299
    },
    {
      "epoch": 6.512185377546944,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 16300
    },
    {
      "epoch": 6.512584898122253,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 16301
    },
    {
      "epoch": 6.512984418697563,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 16302
    },
    {
      "epoch": 6.513383939272872,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 16303
    },
    {
      "epoch": 6.513783459848182,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 16304
    },
    {
      "epoch": 6.514182980423492,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 16305
    },
    {
      "epoch": 6.514582500998801,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 16306
    },
    {
      "epoch": 6.514982021574111,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 16307
    },
    {
      "epoch": 6.515381542149421,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 16308
    },
    {
      "epoch": 6.515781062724731,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 16309
    },
    {
      "epoch": 6.51618058330004,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 16310
    },
    {
      "epoch": 6.5165801038753495,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 16311
    },
    {
      "epoch": 6.516979624450659,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 16312
    },
    {
      "epoch": 6.517379145025969,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 16313
    },
    {
      "epoch": 6.517778665601279,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 16314
    },
    {
      "epoch": 6.518178186176588,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 16315
    },
    {
      "epoch": 6.5185777067518975,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 16316
    },
    {
      "epoch": 6.518977227327207,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 16317
    },
    {
      "epoch": 6.519376747902517,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0637,
      "step": 16318
    },
    {
      "epoch": 6.519776268477827,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 16319
    },
    {
      "epoch": 6.520175789053136,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 16320
    },
    {
      "epoch": 6.520575309628446,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 16321
    },
    {
      "epoch": 6.520974830203755,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 16322
    },
    {
      "epoch": 6.521374350779065,
      "grad_norm": 0.171875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 16323
    },
    {
      "epoch": 6.521773871354375,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 16324
    },
    {
      "epoch": 6.522173391929685,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 16325
    },
    {
      "epoch": 6.5225729125049945,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 16326
    },
    {
      "epoch": 6.522972433080303,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 16327
    },
    {
      "epoch": 6.523371953655613,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 16328
    },
    {
      "epoch": 6.523771474230923,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 16329
    },
    {
      "epoch": 6.524170994806233,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 16330
    },
    {
      "epoch": 6.5245705153815425,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 16331
    },
    {
      "epoch": 6.524970035956851,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 16332
    },
    {
      "epoch": 6.525369556532161,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 16333
    },
    {
      "epoch": 6.525769077107471,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 16334
    },
    {
      "epoch": 6.526168597682781,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 16335
    },
    {
      "epoch": 6.526568118258091,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 16336
    },
    {
      "epoch": 6.5269676388333995,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 16337
    },
    {
      "epoch": 6.527367159408709,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 16338
    },
    {
      "epoch": 6.527766679984019,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 16339
    },
    {
      "epoch": 6.528166200559329,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 16340
    },
    {
      "epoch": 6.528565721134639,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 16341
    },
    {
      "epoch": 6.528965241709948,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 16342
    },
    {
      "epoch": 6.529364762285257,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 16343
    },
    {
      "epoch": 6.529764282860567,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 16344
    },
    {
      "epoch": 6.530163803435877,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 16345
    },
    {
      "epoch": 6.530563324011187,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 16346
    },
    {
      "epoch": 6.5309628445864965,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 16347
    },
    {
      "epoch": 6.531362365161806,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 16348
    },
    {
      "epoch": 6.531761885737115,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 16349
    },
    {
      "epoch": 6.532161406312425,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 16350
    },
    {
      "epoch": 6.532560926887735,
      "grad_norm": 0.171875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 16351
    },
    {
      "epoch": 6.5329604474630445,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 16352
    },
    {
      "epoch": 6.533359968038354,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 16353
    },
    {
      "epoch": 6.533759488613663,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 16354
    },
    {
      "epoch": 6.534159009188973,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 16355
    },
    {
      "epoch": 6.534558529764283,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 16356
    },
    {
      "epoch": 6.5349580503395925,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 16357
    },
    {
      "epoch": 6.535357570914902,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 16358
    },
    {
      "epoch": 6.535757091490212,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 16359
    },
    {
      "epoch": 6.536156612065521,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 16360
    },
    {
      "epoch": 6.536556132640831,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 16361
    },
    {
      "epoch": 6.536955653216141,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 16362
    },
    {
      "epoch": 6.53735517379145,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 16363
    },
    {
      "epoch": 6.53775469436676,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 16364
    },
    {
      "epoch": 6.53815421494207,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 16365
    },
    {
      "epoch": 6.538553735517379,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 16366
    },
    {
      "epoch": 6.538953256092689,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 16367
    },
    {
      "epoch": 6.539352776667998,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 16368
    },
    {
      "epoch": 6.539752297243308,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 16369
    },
    {
      "epoch": 6.540151817818618,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 16370
    },
    {
      "epoch": 6.540551338393927,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 16371
    },
    {
      "epoch": 6.540950858969237,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 16372
    },
    {
      "epoch": 6.5413503795445465,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 16373
    },
    {
      "epoch": 6.541749900119856,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 16374
    },
    {
      "epoch": 6.542149420695166,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 16375
    },
    {
      "epoch": 6.542548941270476,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 16376
    },
    {
      "epoch": 6.542948461845785,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 16377
    },
    {
      "epoch": 6.5433479824210945,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 16378
    },
    {
      "epoch": 6.543747502996404,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 16379
    },
    {
      "epoch": 6.544147023571714,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 16380
    },
    {
      "epoch": 6.544546544147024,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 16381
    },
    {
      "epoch": 6.544946064722334,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 16382
    },
    {
      "epoch": 6.5453455852976425,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 16383
    },
    {
      "epoch": 6.545745105872952,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 16384
    },
    {
      "epoch": 6.546144626448262,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 16385
    },
    {
      "epoch": 6.546544147023572,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 16386
    },
    {
      "epoch": 6.546943667598882,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 16387
    },
    {
      "epoch": 6.547343188174191,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 16388
    },
    {
      "epoch": 6.5477427087495,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 16389
    },
    {
      "epoch": 6.54814222932481,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 16390
    },
    {
      "epoch": 6.54854174990012,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 16391
    },
    {
      "epoch": 6.54894127047543,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 16392
    },
    {
      "epoch": 6.549340791050739,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 16393
    },
    {
      "epoch": 6.549740311626048,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 16394
    },
    {
      "epoch": 6.550139832201358,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 16395
    },
    {
      "epoch": 6.550539352776668,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 16396
    },
    {
      "epoch": 6.550938873351978,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0599,
      "step": 16397
    },
    {
      "epoch": 6.551338393927288,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 16398
    },
    {
      "epoch": 6.551737914502597,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 16399
    },
    {
      "epoch": 6.552137435077906,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 16400
    },
    {
      "epoch": 6.552536955653216,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 16401
    },
    {
      "epoch": 6.552936476228526,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 16402
    },
    {
      "epoch": 6.553335996803836,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 16403
    },
    {
      "epoch": 6.553735517379145,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 16404
    },
    {
      "epoch": 6.554135037954454,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 16405
    },
    {
      "epoch": 6.554534558529764,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 16406
    },
    {
      "epoch": 6.554934079105074,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 16407
    },
    {
      "epoch": 6.555333599680384,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 16408
    },
    {
      "epoch": 6.555733120255693,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 16409
    },
    {
      "epoch": 6.556132640831002,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 16410
    },
    {
      "epoch": 6.556532161406312,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 16411
    },
    {
      "epoch": 6.556931681981622,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 16412
    },
    {
      "epoch": 6.557331202556932,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 16413
    },
    {
      "epoch": 6.5577307231322415,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 16414
    },
    {
      "epoch": 6.558130243707551,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 16415
    },
    {
      "epoch": 6.558529764282861,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 16416
    },
    {
      "epoch": 6.55892928485817,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 16417
    },
    {
      "epoch": 6.55932880543348,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 16418
    },
    {
      "epoch": 6.5597283260087895,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 16419
    },
    {
      "epoch": 6.560127846584099,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 16420
    },
    {
      "epoch": 6.560527367159409,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 16421
    },
    {
      "epoch": 6.560926887734718,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 16422
    },
    {
      "epoch": 6.561326408310028,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 16423
    },
    {
      "epoch": 6.561725928885338,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 16424
    },
    {
      "epoch": 6.562125449460647,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 16425
    },
    {
      "epoch": 6.562524970035957,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 16426
    },
    {
      "epoch": 6.562924490611266,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 16427
    },
    {
      "epoch": 6.563324011186576,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 16428
    },
    {
      "epoch": 6.563723531761886,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 16429
    },
    {
      "epoch": 6.564123052337195,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 16430
    },
    {
      "epoch": 6.564522572912505,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 16431
    },
    {
      "epoch": 6.564922093487815,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 16432
    },
    {
      "epoch": 6.565321614063124,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 16433
    },
    {
      "epoch": 6.565721134638434,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 16434
    },
    {
      "epoch": 6.566120655213743,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 16435
    },
    {
      "epoch": 6.566520175789053,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 16436
    },
    {
      "epoch": 6.566919696364363,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 16437
    },
    {
      "epoch": 6.567319216939673,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 16438
    },
    {
      "epoch": 6.567718737514982,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 16439
    },
    {
      "epoch": 6.5681182580902915,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 16440
    },
    {
      "epoch": 6.568517778665601,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 16441
    },
    {
      "epoch": 6.568917299240911,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 16442
    },
    {
      "epoch": 6.569316819816221,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 16443
    },
    {
      "epoch": 6.56971634039153,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 16444
    },
    {
      "epoch": 6.5701158609668395,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 16445
    },
    {
      "epoch": 6.570515381542149,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 16446
    },
    {
      "epoch": 6.570914902117459,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 16447
    },
    {
      "epoch": 6.571314422692769,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 16448
    },
    {
      "epoch": 6.571713943268079,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 16449
    },
    {
      "epoch": 6.572113463843388,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 16450
    },
    {
      "epoch": 6.572512984418697,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 16451
    },
    {
      "epoch": 6.572912504994007,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 16452
    },
    {
      "epoch": 6.573312025569317,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 16453
    },
    {
      "epoch": 6.573711546144627,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 16454
    },
    {
      "epoch": 6.5741110667199365,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 16455
    },
    {
      "epoch": 6.574510587295245,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 16456
    },
    {
      "epoch": 6.574910107870555,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 16457
    },
    {
      "epoch": 6.575309628445865,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 16458
    },
    {
      "epoch": 6.575709149021175,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 16459
    },
    {
      "epoch": 6.5761086695964845,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 16460
    },
    {
      "epoch": 6.5765081901717934,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 16461
    },
    {
      "epoch": 6.576907710747103,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 16462
    },
    {
      "epoch": 6.577307231322413,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 16463
    },
    {
      "epoch": 6.577706751897723,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 16464
    },
    {
      "epoch": 6.578106272473033,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 16465
    },
    {
      "epoch": 6.578505793048342,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 16466
    },
    {
      "epoch": 6.578905313623651,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 16467
    },
    {
      "epoch": 6.579304834198961,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 16468
    },
    {
      "epoch": 6.579704354774271,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 16469
    },
    {
      "epoch": 6.580103875349581,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 16470
    },
    {
      "epoch": 6.58050339592489,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 16471
    },
    {
      "epoch": 6.5809029165002,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 16472
    },
    {
      "epoch": 6.581302437075509,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 16473
    },
    {
      "epoch": 6.581701957650819,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 16474
    },
    {
      "epoch": 6.582101478226129,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 16475
    },
    {
      "epoch": 6.5825009988014385,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 16476
    },
    {
      "epoch": 6.582900519376748,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 16477
    },
    {
      "epoch": 6.583300039952057,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 16478
    },
    {
      "epoch": 6.583699560527367,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 16479
    },
    {
      "epoch": 6.584099081102677,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 16480
    },
    {
      "epoch": 6.5844986016779865,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 16481
    },
    {
      "epoch": 6.584898122253296,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 16482
    },
    {
      "epoch": 6.585297642828605,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 16483
    },
    {
      "epoch": 6.585697163403915,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 16484
    },
    {
      "epoch": 6.586096683979225,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 16485
    },
    {
      "epoch": 6.5864962045545346,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 16486
    },
    {
      "epoch": 6.586895725129844,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 16487
    },
    {
      "epoch": 6.587295245705154,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 16488
    },
    {
      "epoch": 6.587694766280464,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 16489
    },
    {
      "epoch": 6.588094286855773,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 16490
    },
    {
      "epoch": 6.588493807431083,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 16491
    },
    {
      "epoch": 6.588893328006392,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 16492
    },
    {
      "epoch": 6.589292848581702,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 16493
    },
    {
      "epoch": 6.589692369157012,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 16494
    },
    {
      "epoch": 6.590091889732321,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 16495
    },
    {
      "epoch": 6.590491410307631,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 16496
    },
    {
      "epoch": 6.59089093088294,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 16497
    },
    {
      "epoch": 6.59129045145825,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0576,
      "step": 16498
    },
    {
      "epoch": 6.59168997203356,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 16499
    },
    {
      "epoch": 6.592089492608869,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 16500
    },
    {
      "epoch": 6.592489013184179,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0555,
      "step": 16501
    },
    {
      "epoch": 6.5928885337594885,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 16502
    },
    {
      "epoch": 6.593288054334798,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 16503
    },
    {
      "epoch": 6.593687574910108,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 16504
    },
    {
      "epoch": 6.594087095485418,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 16505
    },
    {
      "epoch": 6.594486616060728,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 16506
    },
    {
      "epoch": 6.5948861366360365,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 16507
    },
    {
      "epoch": 6.595285657211346,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 16508
    },
    {
      "epoch": 6.595685177786656,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 16509
    },
    {
      "epoch": 6.596084698361966,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 16510
    },
    {
      "epoch": 6.596484218937276,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 16511
    },
    {
      "epoch": 6.5968837395125846,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 16512
    },
    {
      "epoch": 6.597283260087894,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 16513
    },
    {
      "epoch": 6.597682780663204,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 16514
    },
    {
      "epoch": 6.598082301238514,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 16515
    },
    {
      "epoch": 6.598481821813824,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 16516
    },
    {
      "epoch": 6.598881342389133,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 16517
    },
    {
      "epoch": 6.599280862964442,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 16518
    },
    {
      "epoch": 6.599680383539752,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 16519
    },
    {
      "epoch": 6.600079904115062,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 16520
    },
    {
      "epoch": 6.600479424690372,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 16521
    },
    {
      "epoch": 6.6008789452656815,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 16522
    },
    {
      "epoch": 6.60127846584099,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 16523
    },
    {
      "epoch": 6.6016779864163,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 16524
    },
    {
      "epoch": 6.60207750699161,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 16525
    },
    {
      "epoch": 6.60247702756692,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 16526
    },
    {
      "epoch": 6.60287654814223,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 16527
    },
    {
      "epoch": 6.603276068717539,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 16528
    },
    {
      "epoch": 6.603675589292848,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 16529
    },
    {
      "epoch": 6.604075109868158,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 16530
    },
    {
      "epoch": 6.604474630443468,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 16531
    },
    {
      "epoch": 6.604874151018778,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 16532
    },
    {
      "epoch": 6.605273671594087,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 16533
    },
    {
      "epoch": 6.605673192169396,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 16534
    },
    {
      "epoch": 6.606072712744706,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 16535
    },
    {
      "epoch": 6.606472233320016,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 16536
    },
    {
      "epoch": 6.606871753895326,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 16537
    },
    {
      "epoch": 6.6072712744706354,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 16538
    },
    {
      "epoch": 6.607670795045945,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 16539
    },
    {
      "epoch": 6.608070315621254,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 16540
    },
    {
      "epoch": 6.608469836196564,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 16541
    },
    {
      "epoch": 6.608869356771874,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 16542
    },
    {
      "epoch": 6.6092688773471835,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 16543
    },
    {
      "epoch": 6.609668397922493,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 16544
    },
    {
      "epoch": 6.610067918497803,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 16545
    },
    {
      "epoch": 6.610467439073112,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 16546
    },
    {
      "epoch": 6.610866959648422,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 16547
    },
    {
      "epoch": 6.6112664802237315,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 16548
    },
    {
      "epoch": 6.611666000799041,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 16549
    },
    {
      "epoch": 6.612065521374351,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 16550
    },
    {
      "epoch": 6.61246504194966,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 16551
    },
    {
      "epoch": 6.61286456252497,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 16552
    },
    {
      "epoch": 6.61326408310028,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0658,
      "step": 16553
    },
    {
      "epoch": 6.613663603675589,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 16554
    },
    {
      "epoch": 6.614063124250899,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 16555
    },
    {
      "epoch": 6.614462644826209,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 16556
    },
    {
      "epoch": 6.614862165401518,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 16557
    },
    {
      "epoch": 6.615261685976828,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 16558
    },
    {
      "epoch": 6.615661206552137,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 16559
    },
    {
      "epoch": 6.616060727127447,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 16560
    },
    {
      "epoch": 6.616460247702757,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 16561
    },
    {
      "epoch": 6.616859768278067,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 16562
    },
    {
      "epoch": 6.617259288853376,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 16563
    },
    {
      "epoch": 6.6176588094286855,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 16564
    },
    {
      "epoch": 6.618058330003995,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 16565
    },
    {
      "epoch": 6.618457850579305,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 16566
    },
    {
      "epoch": 6.618857371154615,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 16567
    },
    {
      "epoch": 6.619256891729924,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 16568
    },
    {
      "epoch": 6.6196564123052335,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 16569
    },
    {
      "epoch": 6.620055932880543,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 16570
    },
    {
      "epoch": 6.620455453455853,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 16571
    },
    {
      "epoch": 6.620854974031163,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 16572
    },
    {
      "epoch": 6.621254494606472,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 16573
    },
    {
      "epoch": 6.6216540151817815,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 16574
    },
    {
      "epoch": 6.622053535757091,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 16575
    },
    {
      "epoch": 6.622453056332401,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 16576
    },
    {
      "epoch": 6.622852576907711,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 16577
    },
    {
      "epoch": 6.623252097483021,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 16578
    },
    {
      "epoch": 6.6236516180583305,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 16579
    },
    {
      "epoch": 6.624051138633639,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 16580
    },
    {
      "epoch": 6.624450659208949,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 16581
    },
    {
      "epoch": 6.624850179784259,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 16582
    },
    {
      "epoch": 6.625249700359569,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 16583
    },
    {
      "epoch": 6.6256492209348785,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 16584
    },
    {
      "epoch": 6.626048741510187,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 16585
    },
    {
      "epoch": 6.626448262085497,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 16586
    },
    {
      "epoch": 6.626847782660807,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 16587
    },
    {
      "epoch": 6.627247303236117,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 16588
    },
    {
      "epoch": 6.627646823811427,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 16589
    },
    {
      "epoch": 6.6280463443867355,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 16590
    },
    {
      "epoch": 6.628445864962045,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 16591
    },
    {
      "epoch": 6.628845385537355,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 16592
    },
    {
      "epoch": 6.629244906112665,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 16593
    },
    {
      "epoch": 6.629644426687975,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 16594
    },
    {
      "epoch": 6.630043947263284,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 16595
    },
    {
      "epoch": 6.630443467838594,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 16596
    },
    {
      "epoch": 6.630842988413903,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 16597
    },
    {
      "epoch": 6.631242508989213,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 16598
    },
    {
      "epoch": 6.631642029564523,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 16599
    },
    {
      "epoch": 6.632041550139832,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 16600
    },
    {
      "epoch": 6.632441070715142,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 16601
    },
    {
      "epoch": 6.632840591290451,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 16602
    },
    {
      "epoch": 6.633240111865761,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 16603
    },
    {
      "epoch": 6.633639632441071,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 16604
    },
    {
      "epoch": 6.6340391530163805,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 16605
    },
    {
      "epoch": 6.63443867359169,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 16606
    },
    {
      "epoch": 6.634838194166999,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 16607
    },
    {
      "epoch": 6.635237714742309,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 16608
    },
    {
      "epoch": 6.635637235317619,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 16609
    },
    {
      "epoch": 6.6360367558929285,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 16610
    },
    {
      "epoch": 6.636436276468238,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 16611
    },
    {
      "epoch": 6.636835797043548,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 16612
    },
    {
      "epoch": 6.637235317618857,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 16613
    },
    {
      "epoch": 6.637634838194167,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 16614
    },
    {
      "epoch": 6.638034358769477,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 16615
    },
    {
      "epoch": 6.638433879344786,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 16616
    },
    {
      "epoch": 6.638833399920096,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 16617
    },
    {
      "epoch": 6.639232920495406,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 16618
    },
    {
      "epoch": 6.639632441070715,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 16619
    },
    {
      "epoch": 6.640031961646025,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 16620
    },
    {
      "epoch": 6.640431482221334,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 16621
    },
    {
      "epoch": 6.640831002796644,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 16622
    },
    {
      "epoch": 6.641230523371954,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 16623
    },
    {
      "epoch": 6.641630043947263,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 16624
    },
    {
      "epoch": 6.642029564522573,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 16625
    },
    {
      "epoch": 6.642429085097882,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 16626
    },
    {
      "epoch": 6.642828605673192,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 16627
    },
    {
      "epoch": 6.643228126248502,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 16628
    },
    {
      "epoch": 6.643627646823812,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 16629
    },
    {
      "epoch": 6.644027167399121,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 16630
    },
    {
      "epoch": 6.6444266879744305,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 16631
    },
    {
      "epoch": 6.64482620854974,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 16632
    },
    {
      "epoch": 6.64522572912505,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 16633
    },
    {
      "epoch": 6.64562524970036,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 16634
    },
    {
      "epoch": 6.64602477027567,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 16635
    },
    {
      "epoch": 6.6464242908509785,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 16636
    },
    {
      "epoch": 6.646823811426288,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 16637
    },
    {
      "epoch": 6.647223332001598,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 16638
    },
    {
      "epoch": 6.647622852576908,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 16639
    },
    {
      "epoch": 6.648022373152218,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 16640
    },
    {
      "epoch": 6.648421893727527,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 16641
    },
    {
      "epoch": 6.648821414302836,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 16642
    },
    {
      "epoch": 6.649220934878146,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 16643
    },
    {
      "epoch": 6.649620455453456,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 16644
    },
    {
      "epoch": 6.650019976028766,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 16645
    },
    {
      "epoch": 6.6504194966040755,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 16646
    },
    {
      "epoch": 6.650819017179384,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 16647
    },
    {
      "epoch": 6.651218537754694,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 16648
    },
    {
      "epoch": 6.651618058330004,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 16649
    },
    {
      "epoch": 6.652017578905314,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 16650
    },
    {
      "epoch": 6.6524170994806235,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 16651
    },
    {
      "epoch": 6.652816620055933,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 16652
    },
    {
      "epoch": 6.653216140631242,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 16653
    },
    {
      "epoch": 6.653615661206552,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 16654
    },
    {
      "epoch": 6.654015181781862,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 16655
    },
    {
      "epoch": 6.654414702357172,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 16656
    },
    {
      "epoch": 6.654814222932481,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 16657
    },
    {
      "epoch": 6.65521374350779,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 16658
    },
    {
      "epoch": 6.6556132640831,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 16659
    },
    {
      "epoch": 6.65601278465841,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 16660
    },
    {
      "epoch": 6.65641230523372,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 16661
    },
    {
      "epoch": 6.656811825809029,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 16662
    },
    {
      "epoch": 6.657211346384338,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 16663
    },
    {
      "epoch": 6.657610866959648,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 16664
    },
    {
      "epoch": 6.658010387534958,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 16665
    },
    {
      "epoch": 6.658409908110268,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 16666
    },
    {
      "epoch": 6.6588094286855775,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 16667
    },
    {
      "epoch": 6.659208949260887,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 16668
    },
    {
      "epoch": 6.659608469836197,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 16669
    },
    {
      "epoch": 6.660007990411506,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 16670
    },
    {
      "epoch": 6.660407510986816,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 16671
    },
    {
      "epoch": 6.6608070315621255,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 16672
    },
    {
      "epoch": 6.661206552137435,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 16673
    },
    {
      "epoch": 6.661606072712745,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 16674
    },
    {
      "epoch": 6.662005593288054,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 16675
    },
    {
      "epoch": 6.662405113863364,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 16676
    },
    {
      "epoch": 6.6628046344386735,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 16677
    },
    {
      "epoch": 6.663204155013983,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 16678
    },
    {
      "epoch": 6.663603675589293,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 16679
    },
    {
      "epoch": 6.664003196164602,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 16680
    },
    {
      "epoch": 6.664402716739912,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 16681
    },
    {
      "epoch": 6.664802237315222,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 16682
    },
    {
      "epoch": 6.665201757890531,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 16683
    },
    {
      "epoch": 6.665601278465841,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 16684
    },
    {
      "epoch": 6.666000799041151,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 16685
    },
    {
      "epoch": 6.666400319616461,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 16686
    },
    {
      "epoch": 6.66679984019177,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 16687
    },
    {
      "epoch": 6.667199360767079,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 16688
    },
    {
      "epoch": 6.667598881342389,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 16689
    },
    {
      "epoch": 6.667998401917699,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 16690
    },
    {
      "epoch": 6.668397922493009,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 16691
    },
    {
      "epoch": 6.668797443068318,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 16692
    },
    {
      "epoch": 6.6691969636436275,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 16693
    },
    {
      "epoch": 6.669596484218937,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 16694
    },
    {
      "epoch": 6.669996004794247,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 16695
    },
    {
      "epoch": 6.670395525369557,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 16696
    },
    {
      "epoch": 6.670795045944866,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 16697
    },
    {
      "epoch": 6.6711945665201755,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 16698
    },
    {
      "epoch": 6.671594087095485,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 16699
    },
    {
      "epoch": 6.671993607670795,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 16700
    },
    {
      "epoch": 6.672393128246105,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 16701
    },
    {
      "epoch": 6.672792648821415,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 16702
    },
    {
      "epoch": 6.6731921693967235,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 16703
    },
    {
      "epoch": 6.673591689972033,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 16704
    },
    {
      "epoch": 6.673991210547343,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 16705
    },
    {
      "epoch": 6.674390731122653,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 16706
    },
    {
      "epoch": 6.674790251697963,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 16707
    },
    {
      "epoch": 6.6751897722732725,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 16708
    },
    {
      "epoch": 6.675589292848581,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 16709
    },
    {
      "epoch": 6.675988813423891,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 16710
    },
    {
      "epoch": 6.676388333999201,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 16711
    },
    {
      "epoch": 6.676787854574511,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 16712
    },
    {
      "epoch": 6.6771873751498205,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 16713
    },
    {
      "epoch": 6.677586895725129,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 16714
    },
    {
      "epoch": 6.677986416300439,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 16715
    },
    {
      "epoch": 6.678385936875749,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 16716
    },
    {
      "epoch": 6.678785457451059,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 16717
    },
    {
      "epoch": 6.679184978026369,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 16718
    },
    {
      "epoch": 6.679584498601678,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 16719
    },
    {
      "epoch": 6.679984019176987,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 16720
    },
    {
      "epoch": 6.680383539752297,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 16721
    },
    {
      "epoch": 6.680783060327607,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 16722
    },
    {
      "epoch": 6.681182580902917,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 16723
    },
    {
      "epoch": 6.681582101478226,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 16724
    },
    {
      "epoch": 6.681981622053536,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 16725
    },
    {
      "epoch": 6.682381142628845,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 16726
    },
    {
      "epoch": 6.682780663204155,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 16727
    },
    {
      "epoch": 6.683180183779465,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 16728
    },
    {
      "epoch": 6.683579704354774,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 16729
    },
    {
      "epoch": 6.683979224930084,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 16730
    },
    {
      "epoch": 6.684378745505393,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 16731
    },
    {
      "epoch": 6.684778266080703,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 16732
    },
    {
      "epoch": 6.685177786656013,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 16733
    },
    {
      "epoch": 6.6855773072313225,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 16734
    },
    {
      "epoch": 6.685976827806632,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 16735
    },
    {
      "epoch": 6.686376348381942,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 16736
    },
    {
      "epoch": 6.686775868957251,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 16737
    },
    {
      "epoch": 6.687175389532561,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 16738
    },
    {
      "epoch": 6.6875749101078705,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 16739
    },
    {
      "epoch": 6.68797443068318,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 16740
    },
    {
      "epoch": 6.68837395125849,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 16741
    },
    {
      "epoch": 6.6887734718338,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 16742
    },
    {
      "epoch": 6.689172992409109,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 16743
    },
    {
      "epoch": 6.689572512984419,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 16744
    },
    {
      "epoch": 6.689972033559728,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 16745
    },
    {
      "epoch": 6.690371554135038,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 16746
    },
    {
      "epoch": 6.690771074710348,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 16747
    },
    {
      "epoch": 6.691170595285657,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 16748
    },
    {
      "epoch": 6.691570115860967,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 16749
    },
    {
      "epoch": 6.691969636436276,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 16750
    },
    {
      "epoch": 6.692369157011586,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 16751
    },
    {
      "epoch": 6.692768677586896,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 16752
    },
    {
      "epoch": 6.693168198162205,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 16753
    },
    {
      "epoch": 6.693567718737515,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 16754
    },
    {
      "epoch": 6.6939672393128244,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 16755
    },
    {
      "epoch": 6.694366759888134,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 16756
    },
    {
      "epoch": 6.694766280463444,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 16757
    },
    {
      "epoch": 6.695165801038754,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 16758
    },
    {
      "epoch": 6.695565321614064,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 16759
    },
    {
      "epoch": 6.6959648421893725,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 16760
    },
    {
      "epoch": 6.696364362764682,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 16761
    },
    {
      "epoch": 6.696763883339992,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 16762
    },
    {
      "epoch": 6.697163403915302,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 16763
    },
    {
      "epoch": 6.697562924490612,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 16764
    },
    {
      "epoch": 6.6979624450659205,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 16765
    },
    {
      "epoch": 6.69836196564123,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 16766
    },
    {
      "epoch": 6.69876148621654,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 16767
    },
    {
      "epoch": 6.69916100679185,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 16768
    },
    {
      "epoch": 6.69956052736716,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 16769
    },
    {
      "epoch": 6.699960047942469,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 16770
    },
    {
      "epoch": 6.700359568517778,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 16771
    },
    {
      "epoch": 6.700759089093088,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 16772
    },
    {
      "epoch": 6.701158609668398,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 16773
    },
    {
      "epoch": 6.701558130243708,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 16774
    },
    {
      "epoch": 6.7019576508190175,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 16775
    },
    {
      "epoch": 6.702357171394327,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 16776
    },
    {
      "epoch": 6.702756691969636,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 16777
    },
    {
      "epoch": 6.703156212544946,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 16778
    },
    {
      "epoch": 6.703555733120256,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 16779
    },
    {
      "epoch": 6.7039552536955656,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 16780
    },
    {
      "epoch": 6.704354774270875,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 16781
    },
    {
      "epoch": 6.704754294846184,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 16782
    },
    {
      "epoch": 6.705153815421494,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 16783
    },
    {
      "epoch": 6.705553335996804,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 16784
    },
    {
      "epoch": 6.705952856572114,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 16785
    },
    {
      "epoch": 6.706352377147423,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 16786
    },
    {
      "epoch": 6.706751897722732,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 16787
    },
    {
      "epoch": 6.707151418298042,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 16788
    },
    {
      "epoch": 6.707550938873352,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 16789
    },
    {
      "epoch": 6.707950459448662,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 16790
    },
    {
      "epoch": 6.708349980023971,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 16791
    },
    {
      "epoch": 6.708749500599281,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 16792
    },
    {
      "epoch": 6.70914902117459,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 16793
    },
    {
      "epoch": 6.7095485417499,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 16794
    },
    {
      "epoch": 6.70994806232521,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 16795
    },
    {
      "epoch": 6.7103475829005195,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 16796
    },
    {
      "epoch": 6.710747103475829,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 16797
    },
    {
      "epoch": 6.711146624051139,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 16798
    },
    {
      "epoch": 6.711546144626448,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 16799
    },
    {
      "epoch": 6.711945665201758,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 16800
    },
    {
      "epoch": 6.7123451857770675,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 16801
    },
    {
      "epoch": 6.712744706352377,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 16802
    },
    {
      "epoch": 6.713144226927687,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 16803
    },
    {
      "epoch": 6.713543747502996,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 16804
    },
    {
      "epoch": 6.713943268078306,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 16805
    },
    {
      "epoch": 6.714342788653616,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 16806
    },
    {
      "epoch": 6.714742309228925,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 16807
    },
    {
      "epoch": 6.715141829804235,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0605,
      "step": 16808
    },
    {
      "epoch": 6.715541350379545,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 16809
    },
    {
      "epoch": 6.715940870954854,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 16810
    },
    {
      "epoch": 6.716340391530164,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 16811
    },
    {
      "epoch": 6.716739912105473,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 16812
    },
    {
      "epoch": 6.717139432680783,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 16813
    },
    {
      "epoch": 6.717538953256093,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 16814
    },
    {
      "epoch": 6.717938473831403,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 16815
    },
    {
      "epoch": 6.718337994406712,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 16816
    },
    {
      "epoch": 6.718737514982021,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 16817
    },
    {
      "epoch": 6.719137035557331,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 16818
    },
    {
      "epoch": 6.719536556132641,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 16819
    },
    {
      "epoch": 6.719936076707951,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 16820
    },
    {
      "epoch": 6.72033559728326,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 16821
    },
    {
      "epoch": 6.7207351178585695,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 16822
    },
    {
      "epoch": 6.721134638433879,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 16823
    },
    {
      "epoch": 6.721534159009189,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 16824
    },
    {
      "epoch": 6.721933679584499,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 16825
    },
    {
      "epoch": 6.722333200159808,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 16826
    },
    {
      "epoch": 6.7227327207351175,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 16827
    },
    {
      "epoch": 6.723132241310427,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 16828
    },
    {
      "epoch": 6.723531761885737,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 16829
    },
    {
      "epoch": 6.723931282461047,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 16830
    },
    {
      "epoch": 6.724330803036357,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 16831
    },
    {
      "epoch": 6.7247303236116664,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 16832
    },
    {
      "epoch": 6.725129844186975,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 16833
    },
    {
      "epoch": 6.725529364762285,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 16834
    },
    {
      "epoch": 6.725928885337595,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 16835
    },
    {
      "epoch": 6.726328405912905,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 16836
    },
    {
      "epoch": 6.7267279264882145,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 16837
    },
    {
      "epoch": 6.727127447063523,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 16838
    },
    {
      "epoch": 6.727526967638833,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 16839
    },
    {
      "epoch": 6.727926488214143,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 16840
    },
    {
      "epoch": 6.728326008789453,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 16841
    },
    {
      "epoch": 6.7287255293647625,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 16842
    },
    {
      "epoch": 6.729125049940071,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 16843
    },
    {
      "epoch": 6.729524570515381,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 16844
    },
    {
      "epoch": 6.729924091090691,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 16845
    },
    {
      "epoch": 6.730323611666001,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 16846
    },
    {
      "epoch": 6.730723132241311,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 16847
    },
    {
      "epoch": 6.73112265281662,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 16848
    },
    {
      "epoch": 6.73152217339193,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 16849
    },
    {
      "epoch": 6.731921693967239,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 16850
    },
    {
      "epoch": 6.732321214542549,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 16851
    },
    {
      "epoch": 6.732720735117859,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 16852
    },
    {
      "epoch": 6.733120255693168,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 16853
    },
    {
      "epoch": 6.733519776268478,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 16854
    },
    {
      "epoch": 6.733919296843787,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 16855
    },
    {
      "epoch": 6.734318817419097,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 16856
    },
    {
      "epoch": 6.734718337994407,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 16857
    },
    {
      "epoch": 6.7351178585697165,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 16858
    },
    {
      "epoch": 6.735517379145026,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 16859
    },
    {
      "epoch": 6.735916899720335,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 16860
    },
    {
      "epoch": 6.736316420295645,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 16861
    },
    {
      "epoch": 6.736715940870955,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 16862
    },
    {
      "epoch": 6.7371154614462645,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 16863
    },
    {
      "epoch": 6.737514982021574,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 16864
    },
    {
      "epoch": 6.737914502596884,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 16865
    },
    {
      "epoch": 6.738314023172194,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 16866
    },
    {
      "epoch": 6.738713543747503,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 16867
    },
    {
      "epoch": 6.7391130643228125,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 16868
    },
    {
      "epoch": 6.739512584898122,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 16869
    },
    {
      "epoch": 6.739912105473432,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 16870
    },
    {
      "epoch": 6.740311626048742,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 16871
    },
    {
      "epoch": 6.740711146624051,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 16872
    },
    {
      "epoch": 6.741110667199361,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 16873
    },
    {
      "epoch": 6.74151018777467,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0539,
      "step": 16874
    },
    {
      "epoch": 6.74190970834998,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 16875
    },
    {
      "epoch": 6.74230922892529,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 16876
    },
    {
      "epoch": 6.742708749500599,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 16877
    },
    {
      "epoch": 6.743108270075909,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 16878
    },
    {
      "epoch": 6.743507790651218,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 16879
    },
    {
      "epoch": 6.743907311226528,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 16880
    },
    {
      "epoch": 6.744306831801838,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 16881
    },
    {
      "epoch": 6.744706352377148,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 16882
    },
    {
      "epoch": 6.745105872952457,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 16883
    },
    {
      "epoch": 6.7455053935277665,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 16884
    },
    {
      "epoch": 6.745904914103076,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 16885
    },
    {
      "epoch": 6.746304434678386,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 16886
    },
    {
      "epoch": 6.746703955253696,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 16887
    },
    {
      "epoch": 6.747103475829006,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 16888
    },
    {
      "epoch": 6.7475029964043145,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 16889
    },
    {
      "epoch": 6.747902516979624,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 16890
    },
    {
      "epoch": 6.748302037554934,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 16891
    },
    {
      "epoch": 6.748701558130244,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 16892
    },
    {
      "epoch": 6.749101078705554,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 16893
    },
    {
      "epoch": 6.7495005992808625,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 16894
    },
    {
      "epoch": 6.749900119856172,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 16895
    },
    {
      "epoch": 6.750299640431482,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 16896
    },
    {
      "epoch": 6.750699161006792,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 16897
    },
    {
      "epoch": 6.751098681582102,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 16898
    },
    {
      "epoch": 6.7514982021574115,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 16899
    },
    {
      "epoch": 6.75189772273272,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 16900
    },
    {
      "epoch": 6.75229724330803,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 16901
    },
    {
      "epoch": 6.75269676388334,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 16902
    },
    {
      "epoch": 6.75309628445865,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 16903
    },
    {
      "epoch": 6.7534958050339595,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 16904
    },
    {
      "epoch": 6.753895325609269,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 16905
    },
    {
      "epoch": 6.754294846184578,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 16906
    },
    {
      "epoch": 6.754694366759888,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 16907
    },
    {
      "epoch": 6.755093887335198,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 16908
    },
    {
      "epoch": 6.755493407910508,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 16909
    },
    {
      "epoch": 6.755892928485817,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 16910
    },
    {
      "epoch": 6.756292449061126,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 16911
    },
    {
      "epoch": 6.756691969636436,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 16912
    },
    {
      "epoch": 6.757091490211746,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 16913
    },
    {
      "epoch": 6.757491010787056,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 16914
    },
    {
      "epoch": 6.757890531362365,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 16915
    },
    {
      "epoch": 6.758290051937674,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 16916
    },
    {
      "epoch": 6.758689572512984,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 16917
    },
    {
      "epoch": 6.759089093088294,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 16918
    },
    {
      "epoch": 6.759488613663604,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 16919
    },
    {
      "epoch": 6.759888134238913,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 16920
    },
    {
      "epoch": 6.760287654814223,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 16921
    },
    {
      "epoch": 6.760687175389533,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 16922
    },
    {
      "epoch": 6.761086695964842,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 16923
    },
    {
      "epoch": 6.761486216540152,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 16924
    },
    {
      "epoch": 6.7618857371154615,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 16925
    },
    {
      "epoch": 6.762285257690771,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 16926
    },
    {
      "epoch": 6.762684778266081,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0591,
      "step": 16927
    },
    {
      "epoch": 6.76308429884139,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 16928
    },
    {
      "epoch": 6.7634838194167,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 16929
    },
    {
      "epoch": 6.7638833399920095,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 16930
    },
    {
      "epoch": 6.764282860567319,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 16931
    },
    {
      "epoch": 6.764682381142629,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 16932
    },
    {
      "epoch": 6.765081901717938,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 16933
    },
    {
      "epoch": 6.765481422293248,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 16934
    },
    {
      "epoch": 6.765880942868558,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 16935
    },
    {
      "epoch": 6.766280463443867,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 16936
    },
    {
      "epoch": 6.766679984019177,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 16937
    },
    {
      "epoch": 6.767079504594487,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 16938
    },
    {
      "epoch": 6.767479025169797,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 16939
    },
    {
      "epoch": 6.767878545745106,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 16940
    },
    {
      "epoch": 6.768278066320415,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 16941
    },
    {
      "epoch": 6.768677586895725,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 16942
    },
    {
      "epoch": 6.769077107471035,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 16943
    },
    {
      "epoch": 6.769476628046345,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 16944
    },
    {
      "epoch": 6.769876148621654,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 16945
    },
    {
      "epoch": 6.770275669196963,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 16946
    },
    {
      "epoch": 6.770675189772273,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 16947
    },
    {
      "epoch": 6.771074710347583,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 16948
    },
    {
      "epoch": 6.771474230922893,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 16949
    },
    {
      "epoch": 6.771873751498202,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 16950
    },
    {
      "epoch": 6.7722732720735115,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 16951
    },
    {
      "epoch": 6.772672792648821,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 16952
    },
    {
      "epoch": 6.773072313224131,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 16953
    },
    {
      "epoch": 6.773471833799441,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 16954
    },
    {
      "epoch": 6.773871354374751,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 16955
    },
    {
      "epoch": 6.77427087495006,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 16956
    },
    {
      "epoch": 6.774670395525369,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 16957
    },
    {
      "epoch": 6.775069916100679,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 16958
    },
    {
      "epoch": 6.775469436675989,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 16959
    },
    {
      "epoch": 6.775868957251299,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 16960
    },
    {
      "epoch": 6.7762684778266085,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 16961
    },
    {
      "epoch": 6.776667998401917,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 16962
    },
    {
      "epoch": 6.777067518977227,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 16963
    },
    {
      "epoch": 6.777467039552537,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 16964
    },
    {
      "epoch": 6.777866560127847,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 16965
    },
    {
      "epoch": 6.7782660807031565,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 16966
    },
    {
      "epoch": 6.778665601278465,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 16967
    },
    {
      "epoch": 6.779065121853775,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 16968
    },
    {
      "epoch": 6.779464642429085,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 16969
    },
    {
      "epoch": 6.779864163004395,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0543,
      "step": 16970
    },
    {
      "epoch": 6.7802636835797045,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 16971
    },
    {
      "epoch": 6.780663204155014,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 16972
    },
    {
      "epoch": 6.781062724730323,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 16973
    },
    {
      "epoch": 6.781462245305633,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 16974
    },
    {
      "epoch": 6.781861765880943,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 16975
    },
    {
      "epoch": 6.782261286456253,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 16976
    },
    {
      "epoch": 6.782660807031562,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 16977
    },
    {
      "epoch": 6.783060327606872,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 16978
    },
    {
      "epoch": 6.783459848182181,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 16979
    },
    {
      "epoch": 6.783859368757491,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 16980
    },
    {
      "epoch": 6.784258889332801,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 16981
    },
    {
      "epoch": 6.78465840990811,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 16982
    },
    {
      "epoch": 6.78505793048342,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 16983
    },
    {
      "epoch": 6.785457451058729,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 16984
    },
    {
      "epoch": 6.785856971634039,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 16985
    },
    {
      "epoch": 6.786256492209349,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 16986
    },
    {
      "epoch": 6.7866560127846585,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 16987
    },
    {
      "epoch": 6.787055533359968,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 16988
    },
    {
      "epoch": 6.787455053935278,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 16989
    },
    {
      "epoch": 6.787854574510587,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 16990
    },
    {
      "epoch": 6.788254095085897,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 16991
    },
    {
      "epoch": 6.7886536156612065,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 16992
    },
    {
      "epoch": 6.789053136236516,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 16993
    },
    {
      "epoch": 6.789452656811826,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 16994
    },
    {
      "epoch": 6.789852177387136,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 16995
    },
    {
      "epoch": 6.790251697962445,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 16996
    },
    {
      "epoch": 6.7906512185377546,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 16997
    },
    {
      "epoch": 6.791050739113064,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 16998
    },
    {
      "epoch": 6.791450259688374,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 16999
    },
    {
      "epoch": 6.791849780263684,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 17000
    },
    {
      "epoch": 6.792249300838993,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 17001
    },
    {
      "epoch": 6.792648821414303,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 17002
    },
    {
      "epoch": 6.793048341989612,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 17003
    },
    {
      "epoch": 6.793447862564922,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 17004
    },
    {
      "epoch": 6.793847383140232,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 17005
    },
    {
      "epoch": 6.794246903715541,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 17006
    },
    {
      "epoch": 6.794646424290851,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 17007
    },
    {
      "epoch": 6.79504594486616,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 17008
    },
    {
      "epoch": 6.79544546544147,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 17009
    },
    {
      "epoch": 6.79584498601678,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 17010
    },
    {
      "epoch": 6.79624450659209,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 17011
    },
    {
      "epoch": 6.7966440271674,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 17012
    },
    {
      "epoch": 6.7970435477427085,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 17013
    },
    {
      "epoch": 6.797443068318018,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 17014
    },
    {
      "epoch": 6.797842588893328,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 17015
    },
    {
      "epoch": 6.798242109468638,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 17016
    },
    {
      "epoch": 6.798641630043948,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 17017
    },
    {
      "epoch": 6.7990411506192565,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 17018
    },
    {
      "epoch": 6.799440671194566,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 17019
    },
    {
      "epoch": 6.799840191769876,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 17020
    },
    {
      "epoch": 6.800239712345186,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 17021
    },
    {
      "epoch": 6.800639232920496,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 17022
    },
    {
      "epoch": 6.801038753495805,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 17023
    },
    {
      "epoch": 6.801438274071114,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 17024
    },
    {
      "epoch": 6.801837794646424,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 17025
    },
    {
      "epoch": 6.802237315221734,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 17026
    },
    {
      "epoch": 6.802636835797044,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 17027
    },
    {
      "epoch": 6.8030363563723535,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 17028
    },
    {
      "epoch": 6.803435876947663,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 17029
    },
    {
      "epoch": 6.803835397522972,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 17030
    },
    {
      "epoch": 6.804234918098282,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 17031
    },
    {
      "epoch": 6.804634438673592,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 17032
    },
    {
      "epoch": 6.8050339592489015,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 17033
    },
    {
      "epoch": 6.805433479824211,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 17034
    },
    {
      "epoch": 6.80583300039952,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 17035
    },
    {
      "epoch": 6.80623252097483,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 17036
    },
    {
      "epoch": 6.80663204155014,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 17037
    },
    {
      "epoch": 6.80703156212545,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 17038
    },
    {
      "epoch": 6.807431082700759,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 17039
    },
    {
      "epoch": 6.807830603276068,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 17040
    },
    {
      "epoch": 6.808230123851378,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 17041
    },
    {
      "epoch": 6.808629644426688,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 17042
    },
    {
      "epoch": 6.809029165001998,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 17043
    },
    {
      "epoch": 6.809428685577307,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 17044
    },
    {
      "epoch": 6.809828206152617,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 17045
    },
    {
      "epoch": 6.810227726727927,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 17046
    },
    {
      "epoch": 6.810627247303236,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 17047
    },
    {
      "epoch": 6.811026767878546,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 17048
    },
    {
      "epoch": 6.8114262884538554,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 17049
    },
    {
      "epoch": 6.811825809029165,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 17050
    },
    {
      "epoch": 6.812225329604475,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 17051
    },
    {
      "epoch": 6.812624850179784,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 17052
    },
    {
      "epoch": 6.813024370755094,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 17053
    },
    {
      "epoch": 6.8134238913304035,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 17054
    },
    {
      "epoch": 6.813823411905713,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 17055
    },
    {
      "epoch": 6.814222932481023,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 17056
    },
    {
      "epoch": 6.814622453056332,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 17057
    },
    {
      "epoch": 6.815021973631642,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 17058
    },
    {
      "epoch": 6.8154214942069515,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 17059
    },
    {
      "epoch": 6.815821014782261,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 17060
    },
    {
      "epoch": 6.816220535357571,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 17061
    },
    {
      "epoch": 6.816620055932881,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 17062
    },
    {
      "epoch": 6.81701957650819,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 17063
    },
    {
      "epoch": 6.8174190970835,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 17064
    },
    {
      "epoch": 6.817818617658809,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 17065
    },
    {
      "epoch": 6.818218138234119,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 17066
    },
    {
      "epoch": 6.818617658809429,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 17067
    },
    {
      "epoch": 6.819017179384739,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 17068
    },
    {
      "epoch": 6.819416699960048,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 17069
    },
    {
      "epoch": 6.819816220535357,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 17070
    },
    {
      "epoch": 6.820215741110667,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 17071
    },
    {
      "epoch": 6.820615261685977,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 17072
    },
    {
      "epoch": 6.821014782261287,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 17073
    },
    {
      "epoch": 6.821414302836596,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 17074
    },
    {
      "epoch": 6.8218138234119055,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 17075
    },
    {
      "epoch": 6.822213343987215,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 17076
    },
    {
      "epoch": 6.822612864562525,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 17077
    },
    {
      "epoch": 6.823012385137835,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 17078
    },
    {
      "epoch": 6.823411905713145,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 17079
    },
    {
      "epoch": 6.8238114262884535,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 17080
    },
    {
      "epoch": 6.824210946863763,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 17081
    },
    {
      "epoch": 6.824610467439073,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 17082
    },
    {
      "epoch": 6.825009988014383,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 17083
    },
    {
      "epoch": 6.825409508589693,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 17084
    },
    {
      "epoch": 6.825809029165002,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 17085
    },
    {
      "epoch": 6.826208549740311,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 17086
    },
    {
      "epoch": 6.826608070315621,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 17087
    },
    {
      "epoch": 6.827007590890931,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 17088
    },
    {
      "epoch": 6.827407111466241,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 17089
    },
    {
      "epoch": 6.8278066320415505,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 17090
    },
    {
      "epoch": 6.828206152616859,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 17091
    },
    {
      "epoch": 6.828605673192169,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 17092
    },
    {
      "epoch": 6.829005193767479,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 17093
    },
    {
      "epoch": 6.829404714342789,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 17094
    },
    {
      "epoch": 6.8298042349180985,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 17095
    },
    {
      "epoch": 6.830203755493407,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 17096
    },
    {
      "epoch": 6.830603276068717,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 17097
    },
    {
      "epoch": 6.831002796644027,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 17098
    },
    {
      "epoch": 6.831402317219337,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 17099
    },
    {
      "epoch": 6.831801837794647,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 17100
    },
    {
      "epoch": 6.832201358369956,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 17101
    },
    {
      "epoch": 6.832600878945266,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 17102
    },
    {
      "epoch": 6.833000399520575,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 17103
    },
    {
      "epoch": 6.833399920095885,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 17104
    },
    {
      "epoch": 6.833799440671195,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 17105
    },
    {
      "epoch": 6.834198961246504,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 17106
    },
    {
      "epoch": 6.834598481821814,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 17107
    },
    {
      "epoch": 6.834998002397123,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 17108
    },
    {
      "epoch": 6.835397522972433,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 17109
    },
    {
      "epoch": 6.835797043547743,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 17110
    },
    {
      "epoch": 6.836196564123052,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 17111
    },
    {
      "epoch": 6.836596084698362,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 17112
    },
    {
      "epoch": 6.836995605273671,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 17113
    },
    {
      "epoch": 6.837395125848981,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 17114
    },
    {
      "epoch": 6.837794646424291,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 17115
    },
    {
      "epoch": 6.8381941669996005,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 17116
    },
    {
      "epoch": 6.83859368757491,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 17117
    },
    {
      "epoch": 6.83899320815022,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 17118
    },
    {
      "epoch": 6.83939272872553,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 17119
    },
    {
      "epoch": 6.839792249300839,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 17120
    },
    {
      "epoch": 6.8401917698761485,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 17121
    },
    {
      "epoch": 6.840591290451458,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 17122
    },
    {
      "epoch": 6.840990811026768,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 17123
    },
    {
      "epoch": 6.841390331602078,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 17124
    },
    {
      "epoch": 6.841789852177387,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 17125
    },
    {
      "epoch": 6.842189372752697,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 17126
    },
    {
      "epoch": 6.842588893328006,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 17127
    },
    {
      "epoch": 6.842988413903316,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 17128
    },
    {
      "epoch": 6.843387934478626,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 17129
    },
    {
      "epoch": 6.843787455053935,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 17130
    },
    {
      "epoch": 6.844186975629245,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 17131
    },
    {
      "epoch": 6.844586496204554,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 17132
    },
    {
      "epoch": 6.844986016779864,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 17133
    },
    {
      "epoch": 6.845385537355174,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 17134
    },
    {
      "epoch": 6.845785057930484,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 17135
    },
    {
      "epoch": 6.8461845785057935,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 17136
    },
    {
      "epoch": 6.846584099081102,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 17137
    },
    {
      "epoch": 6.846983619656412,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 17138
    },
    {
      "epoch": 6.847383140231722,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 17139
    },
    {
      "epoch": 6.847782660807032,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 17140
    },
    {
      "epoch": 6.848182181382342,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 17141
    },
    {
      "epoch": 6.8485817019576505,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 17142
    },
    {
      "epoch": 6.84898122253296,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 17143
    },
    {
      "epoch": 6.84938074310827,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 17144
    },
    {
      "epoch": 6.84978026368358,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 17145
    },
    {
      "epoch": 6.85017978425889,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 17146
    },
    {
      "epoch": 6.8505793048341985,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 17147
    },
    {
      "epoch": 6.850978825409508,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 17148
    },
    {
      "epoch": 6.851378345984818,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 17149
    },
    {
      "epoch": 6.851777866560128,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 17150
    },
    {
      "epoch": 6.852177387135438,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 17151
    },
    {
      "epoch": 6.8525769077107475,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 17152
    },
    {
      "epoch": 6.852976428286056,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 17153
    },
    {
      "epoch": 6.853375948861366,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 17154
    },
    {
      "epoch": 6.853775469436676,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 17155
    },
    {
      "epoch": 6.854174990011986,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 17156
    },
    {
      "epoch": 6.8545745105872955,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 17157
    },
    {
      "epoch": 6.854974031162605,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 17158
    },
    {
      "epoch": 6.855373551737914,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 17159
    },
    {
      "epoch": 6.855773072313224,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 17160
    },
    {
      "epoch": 6.856172592888534,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 17161
    },
    {
      "epoch": 6.8565721134638435,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 17162
    },
    {
      "epoch": 6.856971634039153,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 17163
    },
    {
      "epoch": 6.857371154614462,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 17164
    },
    {
      "epoch": 6.857770675189772,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0569,
      "step": 17165
    },
    {
      "epoch": 6.858170195765082,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 17166
    },
    {
      "epoch": 6.858569716340392,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 17167
    },
    {
      "epoch": 6.858969236915701,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 17168
    },
    {
      "epoch": 6.859368757491011,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 17169
    },
    {
      "epoch": 6.85976827806632,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 17170
    },
    {
      "epoch": 6.86016779864163,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 17171
    },
    {
      "epoch": 6.86056731921694,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 17172
    },
    {
      "epoch": 6.860966839792249,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 17173
    },
    {
      "epoch": 6.861366360367559,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 17174
    },
    {
      "epoch": 6.861765880942869,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 17175
    },
    {
      "epoch": 6.862165401518178,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 17176
    },
    {
      "epoch": 6.862564922093488,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 17177
    },
    {
      "epoch": 6.8629644426687975,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 17178
    },
    {
      "epoch": 6.863363963244107,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 17179
    },
    {
      "epoch": 6.863763483819417,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 17180
    },
    {
      "epoch": 6.864163004394726,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 17181
    },
    {
      "epoch": 6.864562524970036,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 17182
    },
    {
      "epoch": 6.8649620455453455,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 17183
    },
    {
      "epoch": 6.865361566120655,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 17184
    },
    {
      "epoch": 6.865761086695965,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 17185
    },
    {
      "epoch": 6.866160607271274,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 17186
    },
    {
      "epoch": 6.866560127846584,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 17187
    },
    {
      "epoch": 6.8669596484218935,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 17188
    },
    {
      "epoch": 6.867359168997203,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 17189
    },
    {
      "epoch": 6.867758689572513,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 17190
    },
    {
      "epoch": 6.868158210147823,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 17191
    },
    {
      "epoch": 6.868557730723133,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 17192
    },
    {
      "epoch": 6.868957251298442,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 17193
    },
    {
      "epoch": 6.869356771873751,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 17194
    },
    {
      "epoch": 6.869756292449061,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 17195
    },
    {
      "epoch": 6.870155813024371,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 17196
    },
    {
      "epoch": 6.870555333599681,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 17197
    },
    {
      "epoch": 6.87095485417499,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 17198
    },
    {
      "epoch": 6.871354374750299,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 17199
    },
    {
      "epoch": 6.871753895325609,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 17200
    },
    {
      "epoch": 6.872153415900919,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 17201
    },
    {
      "epoch": 6.872552936476229,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 17202
    },
    {
      "epoch": 6.872952457051538,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 17203
    },
    {
      "epoch": 6.8733519776268475,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 17204
    },
    {
      "epoch": 6.873751498202157,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 17205
    },
    {
      "epoch": 6.874151018777467,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 17206
    },
    {
      "epoch": 6.874550539352777,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 17207
    },
    {
      "epoch": 6.874950059928087,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 17208
    },
    {
      "epoch": 6.875349580503396,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 17209
    },
    {
      "epoch": 6.875749101078705,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 17210
    },
    {
      "epoch": 6.876148621654015,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 17211
    },
    {
      "epoch": 6.876548142229325,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 17212
    },
    {
      "epoch": 6.876947662804635,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 17213
    },
    {
      "epoch": 6.877347183379944,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 17214
    },
    {
      "epoch": 6.877746703955253,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 17215
    },
    {
      "epoch": 6.878146224530563,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 17216
    },
    {
      "epoch": 6.878545745105873,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 17217
    },
    {
      "epoch": 6.878945265681183,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 17218
    },
    {
      "epoch": 6.8793447862564925,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 17219
    },
    {
      "epoch": 6.879744306831801,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 17220
    },
    {
      "epoch": 6.880143827407111,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 17221
    },
    {
      "epoch": 6.880543347982421,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 17222
    },
    {
      "epoch": 6.880942868557731,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 17223
    },
    {
      "epoch": 6.8813423891330405,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 17224
    },
    {
      "epoch": 6.88174190970835,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 17225
    },
    {
      "epoch": 6.88214143028366,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 17226
    },
    {
      "epoch": 6.882540950858969,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 17227
    },
    {
      "epoch": 6.882940471434279,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 17228
    },
    {
      "epoch": 6.883339992009589,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 17229
    },
    {
      "epoch": 6.883739512584898,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 17230
    },
    {
      "epoch": 6.884139033160208,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 17231
    },
    {
      "epoch": 6.884538553735517,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 17232
    },
    {
      "epoch": 6.884938074310827,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 17233
    },
    {
      "epoch": 6.885337594886137,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 17234
    },
    {
      "epoch": 6.885737115461446,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 17235
    },
    {
      "epoch": 6.886136636036756,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 17236
    },
    {
      "epoch": 6.886536156612065,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 17237
    },
    {
      "epoch": 6.886935677187375,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 17238
    },
    {
      "epoch": 6.887335197762685,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 17239
    },
    {
      "epoch": 6.887734718337994,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 17240
    },
    {
      "epoch": 6.888134238913304,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 17241
    },
    {
      "epoch": 6.888533759488614,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 17242
    },
    {
      "epoch": 6.888933280063923,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 17243
    },
    {
      "epoch": 6.889332800639233,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 17244
    },
    {
      "epoch": 6.8897323212145425,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 17245
    },
    {
      "epoch": 6.890131841789852,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 17246
    },
    {
      "epoch": 6.890531362365162,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 17247
    },
    {
      "epoch": 6.890930882940472,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 17248
    },
    {
      "epoch": 6.891330403515781,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 17249
    },
    {
      "epoch": 6.8917299240910905,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 17250
    },
    {
      "epoch": 6.8921294446664,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 17251
    },
    {
      "epoch": 6.89252896524171,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 17252
    },
    {
      "epoch": 6.89292848581702,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 17253
    },
    {
      "epoch": 6.893328006392329,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 17254
    },
    {
      "epoch": 6.893727526967639,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 17255
    },
    {
      "epoch": 6.894127047542948,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 17256
    },
    {
      "epoch": 6.894526568118258,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 17257
    },
    {
      "epoch": 6.894926088693568,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 17258
    },
    {
      "epoch": 6.895325609268878,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 17259
    },
    {
      "epoch": 6.895725129844187,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 17260
    },
    {
      "epoch": 6.896124650419496,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 17261
    },
    {
      "epoch": 6.896524170994806,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 17262
    },
    {
      "epoch": 6.896923691570116,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 17263
    },
    {
      "epoch": 6.897323212145426,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 17264
    },
    {
      "epoch": 6.8977227327207356,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 17265
    },
    {
      "epoch": 6.8981222532960444,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 17266
    },
    {
      "epoch": 6.898521773871354,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 17267
    },
    {
      "epoch": 6.898921294446664,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 17268
    },
    {
      "epoch": 6.899320815021974,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 17269
    },
    {
      "epoch": 6.899720335597284,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 17270
    },
    {
      "epoch": 6.9001198561725925,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 17271
    },
    {
      "epoch": 6.900519376747902,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 17272
    },
    {
      "epoch": 6.900918897323212,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 17273
    },
    {
      "epoch": 6.901318417898522,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 17274
    },
    {
      "epoch": 6.901717938473832,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 17275
    },
    {
      "epoch": 6.9021174590491405,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 17276
    },
    {
      "epoch": 6.90251697962445,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 17277
    },
    {
      "epoch": 6.90291650019976,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 17278
    },
    {
      "epoch": 6.90331602077507,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 17279
    },
    {
      "epoch": 6.90371554135038,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 17280
    },
    {
      "epoch": 6.9041150619256895,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 17281
    },
    {
      "epoch": 6.904514582500999,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 17282
    },
    {
      "epoch": 6.904914103076308,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 17283
    },
    {
      "epoch": 6.905313623651618,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 17284
    },
    {
      "epoch": 6.905713144226928,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 17285
    },
    {
      "epoch": 6.9061126648022375,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 17286
    },
    {
      "epoch": 6.906512185377547,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 17287
    },
    {
      "epoch": 6.906911705952856,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 17288
    },
    {
      "epoch": 6.907311226528166,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 17289
    },
    {
      "epoch": 6.907710747103476,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 17290
    },
    {
      "epoch": 6.9081102676787856,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 17291
    },
    {
      "epoch": 6.908509788254095,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 17292
    },
    {
      "epoch": 6.908909308829404,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 17293
    },
    {
      "epoch": 6.909308829404714,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 17294
    },
    {
      "epoch": 6.909708349980024,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 17295
    },
    {
      "epoch": 6.910107870555334,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 17296
    },
    {
      "epoch": 6.910507391130643,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 17297
    },
    {
      "epoch": 6.910906911705953,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 17298
    },
    {
      "epoch": 6.911306432281263,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 17299
    },
    {
      "epoch": 6.911705952856572,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 17300
    },
    {
      "epoch": 6.912105473431882,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 17301
    },
    {
      "epoch": 6.912504994007191,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 17302
    },
    {
      "epoch": 6.912904514582501,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 17303
    },
    {
      "epoch": 6.913304035157811,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 17304
    },
    {
      "epoch": 6.91370355573312,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 17305
    },
    {
      "epoch": 6.91410307630843,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 17306
    },
    {
      "epoch": 6.9145025968837395,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 17307
    },
    {
      "epoch": 6.914902117459049,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 17308
    },
    {
      "epoch": 6.915301638034359,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 17309
    },
    {
      "epoch": 6.915701158609668,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 17310
    },
    {
      "epoch": 6.916100679184978,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 17311
    },
    {
      "epoch": 6.9165001997602875,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 17312
    },
    {
      "epoch": 6.916899720335597,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 17313
    },
    {
      "epoch": 6.917299240910907,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 17314
    },
    {
      "epoch": 6.917698761486217,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 17315
    },
    {
      "epoch": 6.918098282061527,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 17316
    },
    {
      "epoch": 6.918497802636836,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 17317
    },
    {
      "epoch": 6.918897323212145,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 17318
    },
    {
      "epoch": 6.919296843787455,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 17319
    },
    {
      "epoch": 6.919696364362765,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 17320
    },
    {
      "epoch": 6.920095884938075,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 17321
    },
    {
      "epoch": 6.920495405513384,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 17322
    },
    {
      "epoch": 6.920894926088693,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 17323
    },
    {
      "epoch": 6.921294446664003,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 17324
    },
    {
      "epoch": 6.921693967239313,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 17325
    },
    {
      "epoch": 6.922093487814623,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 17326
    },
    {
      "epoch": 6.922493008389932,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 17327
    },
    {
      "epoch": 6.922892528965241,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 17328
    },
    {
      "epoch": 6.923292049540551,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 17329
    },
    {
      "epoch": 6.923691570115861,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 17330
    },
    {
      "epoch": 6.924091090691171,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0556,
      "step": 17331
    },
    {
      "epoch": 6.924490611266481,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 17332
    },
    {
      "epoch": 6.9248901318417895,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 17333
    },
    {
      "epoch": 6.925289652417099,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 17334
    },
    {
      "epoch": 6.925689172992409,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 17335
    },
    {
      "epoch": 6.926088693567719,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 17336
    },
    {
      "epoch": 6.926488214143029,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 17337
    },
    {
      "epoch": 6.926887734718338,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 17338
    },
    {
      "epoch": 6.927287255293647,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 17339
    },
    {
      "epoch": 6.927686775868957,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 17340
    },
    {
      "epoch": 6.928086296444267,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 17341
    },
    {
      "epoch": 6.928485817019577,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 17342
    },
    {
      "epoch": 6.9288853375948865,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 17343
    },
    {
      "epoch": 6.929284858170195,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 17344
    },
    {
      "epoch": 6.929684378745505,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 17345
    },
    {
      "epoch": 6.930083899320815,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 17346
    },
    {
      "epoch": 6.930483419896125,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 17347
    },
    {
      "epoch": 6.9308829404714345,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 17348
    },
    {
      "epoch": 6.931282461046744,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 17349
    },
    {
      "epoch": 6.931681981622053,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 17350
    },
    {
      "epoch": 6.932081502197363,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 17351
    },
    {
      "epoch": 6.932481022772673,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 17352
    },
    {
      "epoch": 6.9328805433479825,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 17353
    },
    {
      "epoch": 6.933280063923292,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 17354
    },
    {
      "epoch": 6.933679584498602,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 17355
    },
    {
      "epoch": 6.934079105073911,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 17356
    },
    {
      "epoch": 6.934478625649221,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 17357
    },
    {
      "epoch": 6.934878146224531,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 17358
    },
    {
      "epoch": 6.93527766679984,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 17359
    },
    {
      "epoch": 6.93567718737515,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 17360
    },
    {
      "epoch": 6.936076707950459,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 17361
    },
    {
      "epoch": 6.936476228525769,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 17362
    },
    {
      "epoch": 6.936875749101079,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 17363
    },
    {
      "epoch": 6.937275269676388,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 17364
    },
    {
      "epoch": 6.937674790251698,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 17365
    },
    {
      "epoch": 6.938074310827007,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 17366
    },
    {
      "epoch": 6.938473831402317,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 17367
    },
    {
      "epoch": 6.938873351977627,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 17368
    },
    {
      "epoch": 6.9392728725529365,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 17369
    },
    {
      "epoch": 6.939672393128246,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 17370
    },
    {
      "epoch": 6.940071913703556,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 17371
    },
    {
      "epoch": 6.940471434278866,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 17372
    },
    {
      "epoch": 6.940870954854175,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 17373
    },
    {
      "epoch": 6.9412704754294845,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 17374
    },
    {
      "epoch": 6.941669996004794,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 17375
    },
    {
      "epoch": 6.942069516580104,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 17376
    },
    {
      "epoch": 6.942469037155414,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 17377
    },
    {
      "epoch": 6.942868557730723,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 17378
    },
    {
      "epoch": 6.9432680783060325,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 17379
    },
    {
      "epoch": 6.943667598881342,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 17380
    },
    {
      "epoch": 6.944067119456652,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 17381
    },
    {
      "epoch": 6.944466640031962,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 17382
    },
    {
      "epoch": 6.944866160607271,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 17383
    },
    {
      "epoch": 6.945265681182581,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 17384
    },
    {
      "epoch": 6.94566520175789,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 17385
    },
    {
      "epoch": 6.9460647223332,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 17386
    },
    {
      "epoch": 6.94646424290851,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 17387
    },
    {
      "epoch": 6.94686376348382,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 17388
    },
    {
      "epoch": 6.9472632840591295,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 17389
    },
    {
      "epoch": 6.947662804634438,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 17390
    },
    {
      "epoch": 6.948062325209748,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 17391
    },
    {
      "epoch": 6.948461845785058,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 17392
    },
    {
      "epoch": 6.948861366360368,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 17393
    },
    {
      "epoch": 6.949260886935678,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 17394
    },
    {
      "epoch": 6.9496604075109865,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 17395
    },
    {
      "epoch": 6.950059928086296,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 17396
    },
    {
      "epoch": 6.950459448661606,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 17397
    },
    {
      "epoch": 6.950858969236916,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 17398
    },
    {
      "epoch": 6.951258489812226,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 17399
    },
    {
      "epoch": 6.9516580103875345,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 17400
    },
    {
      "epoch": 6.952057530962844,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 17401
    },
    {
      "epoch": 6.952457051538154,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 17402
    },
    {
      "epoch": 6.952856572113464,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 17403
    },
    {
      "epoch": 6.953256092688774,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 17404
    },
    {
      "epoch": 6.953655613264083,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 17405
    },
    {
      "epoch": 6.954055133839393,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 17406
    },
    {
      "epoch": 6.954454654414702,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 17407
    },
    {
      "epoch": 6.954854174990012,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 17408
    },
    {
      "epoch": 6.955253695565322,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 17409
    },
    {
      "epoch": 6.9556532161406315,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0573,
      "step": 17410
    },
    {
      "epoch": 6.956052736715941,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 17411
    },
    {
      "epoch": 6.95645225729125,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 17412
    },
    {
      "epoch": 6.95685177786656,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 17413
    },
    {
      "epoch": 6.95725129844187,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 17414
    },
    {
      "epoch": 6.9576508190171795,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 17415
    },
    {
      "epoch": 6.958050339592489,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 17416
    },
    {
      "epoch": 6.958449860167798,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 17417
    },
    {
      "epoch": 6.958849380743108,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 17418
    },
    {
      "epoch": 6.959248901318418,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 17419
    },
    {
      "epoch": 6.959648421893728,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 17420
    },
    {
      "epoch": 6.960047942469037,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 17421
    },
    {
      "epoch": 6.960447463044347,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 17422
    },
    {
      "epoch": 6.960846983619656,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 17423
    },
    {
      "epoch": 6.961246504194966,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 17424
    },
    {
      "epoch": 6.961646024770276,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 17425
    },
    {
      "epoch": 6.962045545345585,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 17426
    },
    {
      "epoch": 6.962445065920895,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 17427
    },
    {
      "epoch": 6.962844586496205,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 17428
    },
    {
      "epoch": 6.963244107071514,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 17429
    },
    {
      "epoch": 6.963643627646824,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 17430
    },
    {
      "epoch": 6.964043148222133,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 17431
    },
    {
      "epoch": 6.964442668797443,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 17432
    },
    {
      "epoch": 6.964842189372753,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 17433
    },
    {
      "epoch": 6.965241709948062,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 17434
    },
    {
      "epoch": 6.965641230523372,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 17435
    },
    {
      "epoch": 6.9660407510986815,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 17436
    },
    {
      "epoch": 6.966440271673991,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 17437
    },
    {
      "epoch": 6.966839792249301,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 17438
    },
    {
      "epoch": 6.967239312824611,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 17439
    },
    {
      "epoch": 6.96763883339992,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 17440
    },
    {
      "epoch": 6.9680383539752295,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 17441
    },
    {
      "epoch": 6.968437874550539,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 17442
    },
    {
      "epoch": 6.968837395125849,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 17443
    },
    {
      "epoch": 6.969236915701159,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 17444
    },
    {
      "epoch": 6.969636436276469,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 17445
    },
    {
      "epoch": 6.970035956851778,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 17446
    },
    {
      "epoch": 6.970435477427087,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 17447
    },
    {
      "epoch": 6.970834998002397,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 17448
    },
    {
      "epoch": 6.971234518577707,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 17449
    },
    {
      "epoch": 6.971634039153017,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 17450
    },
    {
      "epoch": 6.972033559728326,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 17451
    },
    {
      "epoch": 6.972433080303635,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 17452
    },
    {
      "epoch": 6.972832600878945,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 17453
    },
    {
      "epoch": 6.973232121454255,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 17454
    },
    {
      "epoch": 6.973631642029565,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 17455
    },
    {
      "epoch": 6.974031162604874,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 17456
    },
    {
      "epoch": 6.974430683180183,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 17457
    },
    {
      "epoch": 6.974830203755493,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 17458
    },
    {
      "epoch": 6.975229724330803,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 17459
    },
    {
      "epoch": 6.975629244906113,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 17460
    },
    {
      "epoch": 6.976028765481423,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 17461
    },
    {
      "epoch": 6.976428286056732,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 17462
    },
    {
      "epoch": 6.976827806632041,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 17463
    },
    {
      "epoch": 6.977227327207351,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 17464
    },
    {
      "epoch": 6.977626847782661,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 17465
    },
    {
      "epoch": 6.978026368357971,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 17466
    },
    {
      "epoch": 6.97842588893328,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 17467
    },
    {
      "epoch": 6.978825409508589,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 17468
    },
    {
      "epoch": 6.979224930083899,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 17469
    },
    {
      "epoch": 6.979624450659209,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 17470
    },
    {
      "epoch": 6.980023971234519,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 17471
    },
    {
      "epoch": 6.9804234918098285,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 17472
    },
    {
      "epoch": 6.980823012385137,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 17473
    },
    {
      "epoch": 6.981222532960447,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 17474
    },
    {
      "epoch": 6.981622053535757,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 17475
    },
    {
      "epoch": 6.982021574111067,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 17476
    },
    {
      "epoch": 6.9824210946863765,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 17477
    },
    {
      "epoch": 6.982820615261686,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 17478
    },
    {
      "epoch": 6.983220135836996,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 17479
    },
    {
      "epoch": 6.983619656412305,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 17480
    },
    {
      "epoch": 6.984019176987615,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 17481
    },
    {
      "epoch": 6.9844186975629245,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 17482
    },
    {
      "epoch": 6.984818218138234,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 17483
    },
    {
      "epoch": 6.985217738713544,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 17484
    },
    {
      "epoch": 6.985617259288853,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 17485
    },
    {
      "epoch": 6.986016779864163,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 17486
    },
    {
      "epoch": 6.986416300439473,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 17487
    },
    {
      "epoch": 6.986815821014782,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 17488
    },
    {
      "epoch": 6.987215341590092,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 17489
    },
    {
      "epoch": 6.987614862165401,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 17490
    },
    {
      "epoch": 6.988014382740711,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 17491
    },
    {
      "epoch": 6.988413903316021,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 17492
    },
    {
      "epoch": 6.98881342389133,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 17493
    },
    {
      "epoch": 6.98921294446664,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 17494
    },
    {
      "epoch": 6.98961246504195,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 17495
    },
    {
      "epoch": 6.99001198561726,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 17496
    },
    {
      "epoch": 6.990411506192569,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 17497
    },
    {
      "epoch": 6.9908110267678785,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 17498
    },
    {
      "epoch": 6.991210547343188,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 17499
    },
    {
      "epoch": 6.991610067918498,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 17500
    },
    {
      "epoch": 6.992009588493808,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 17501
    },
    {
      "epoch": 6.992409109069117,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 17502
    },
    {
      "epoch": 6.9928086296444265,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 17503
    },
    {
      "epoch": 6.993208150219736,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 17504
    },
    {
      "epoch": 6.993607670795046,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 17505
    },
    {
      "epoch": 6.994007191370356,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 17506
    },
    {
      "epoch": 6.994406711945665,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 17507
    },
    {
      "epoch": 6.9948062325209746,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 17508
    },
    {
      "epoch": 6.995205753096284,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 17509
    },
    {
      "epoch": 6.995605273671594,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 17510
    },
    {
      "epoch": 6.996004794246904,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 17511
    },
    {
      "epoch": 6.996404314822214,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 17512
    },
    {
      "epoch": 6.996803835397523,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 17513
    },
    {
      "epoch": 6.997203355972832,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 17514
    },
    {
      "epoch": 6.997602876548142,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 17515
    },
    {
      "epoch": 6.998002397123452,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 17516
    },
    {
      "epoch": 6.998401917698762,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 17517
    },
    {
      "epoch": 6.9988014382740715,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 17518
    },
    {
      "epoch": 6.99920095884938,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 17519
    },
    {
      "epoch": 6.99960047942469,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 17520
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 17521
    },
    {
      "epoch": 7.00039952057531,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 17522
    },
    {
      "epoch": 7.00079904115062,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 17523
    },
    {
      "epoch": 7.0011985617259285,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 17524
    },
    {
      "epoch": 7.001598082301238,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 17525
    },
    {
      "epoch": 7.001997602876548,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 17526
    },
    {
      "epoch": 7.002397123451858,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0613,
      "step": 17527
    },
    {
      "epoch": 7.002796644027168,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 17528
    },
    {
      "epoch": 7.003196164602477,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 17529
    },
    {
      "epoch": 7.003595685177786,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 17530
    },
    {
      "epoch": 7.003995205753096,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 17531
    },
    {
      "epoch": 7.004394726328406,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 17532
    },
    {
      "epoch": 7.004794246903716,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 17533
    },
    {
      "epoch": 7.0051937674790254,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 17534
    },
    {
      "epoch": 7.005593288054335,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 17535
    },
    {
      "epoch": 7.005992808629644,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 17536
    },
    {
      "epoch": 7.006392329204954,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 17537
    },
    {
      "epoch": 7.006791849780264,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 17538
    },
    {
      "epoch": 7.0071913703555735,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 17539
    },
    {
      "epoch": 7.007590890930883,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 17540
    },
    {
      "epoch": 7.007990411506192,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 17541
    },
    {
      "epoch": 7.008389932081502,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 17542
    },
    {
      "epoch": 7.008789452656812,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 17543
    },
    {
      "epoch": 7.0091889732321215,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 17544
    },
    {
      "epoch": 7.009588493807431,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 17545
    },
    {
      "epoch": 7.009988014382741,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 17546
    },
    {
      "epoch": 7.01038753495805,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 17547
    },
    {
      "epoch": 7.01078705553336,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 17548
    },
    {
      "epoch": 7.01118657610867,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 17549
    },
    {
      "epoch": 7.011586096683979,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 17550
    },
    {
      "epoch": 7.011985617259289,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 17551
    },
    {
      "epoch": 7.012385137834598,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 17552
    },
    {
      "epoch": 7.012784658409908,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 17553
    },
    {
      "epoch": 7.013184178985218,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 17554
    },
    {
      "epoch": 7.013583699560527,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 17555
    },
    {
      "epoch": 7.013983220135837,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 17556
    },
    {
      "epoch": 7.014382740711147,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 17557
    },
    {
      "epoch": 7.014782261286456,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 17558
    },
    {
      "epoch": 7.015181781861766,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 17559
    },
    {
      "epoch": 7.0155813024370755,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 17560
    },
    {
      "epoch": 7.015980823012385,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 17561
    },
    {
      "epoch": 7.016380343587695,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 17562
    },
    {
      "epoch": 7.016779864163005,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 17563
    },
    {
      "epoch": 7.017179384738314,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 17564
    },
    {
      "epoch": 7.0175789053136235,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 17565
    },
    {
      "epoch": 7.017978425888933,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 17566
    },
    {
      "epoch": 7.018377946464243,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 17567
    },
    {
      "epoch": 7.018777467039553,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 17568
    },
    {
      "epoch": 7.019176987614862,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 17569
    },
    {
      "epoch": 7.0195765081901715,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 17570
    },
    {
      "epoch": 7.019976028765481,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 17571
    },
    {
      "epoch": 7.020375549340791,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 17572
    },
    {
      "epoch": 7.020775069916101,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 17573
    },
    {
      "epoch": 7.021174590491411,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 17574
    },
    {
      "epoch": 7.02157411106672,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0592,
      "step": 17575
    },
    {
      "epoch": 7.021973631642029,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 17576
    },
    {
      "epoch": 7.022373152217339,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 17577
    },
    {
      "epoch": 7.022772672792649,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 17578
    },
    {
      "epoch": 7.023172193367959,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 17579
    },
    {
      "epoch": 7.0235717139432685,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 17580
    },
    {
      "epoch": 7.023971234518577,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 17581
    },
    {
      "epoch": 7.024370755093887,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 17582
    },
    {
      "epoch": 7.024770275669197,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 17583
    },
    {
      "epoch": 7.025169796244507,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 17584
    },
    {
      "epoch": 7.025569316819817,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 17585
    },
    {
      "epoch": 7.0259688373951255,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 17586
    },
    {
      "epoch": 7.026368357970435,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 17587
    },
    {
      "epoch": 7.026767878545745,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 17588
    },
    {
      "epoch": 7.027167399121055,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 17589
    },
    {
      "epoch": 7.027566919696365,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 17590
    },
    {
      "epoch": 7.027966440271674,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 17591
    },
    {
      "epoch": 7.028365960846983,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 17592
    },
    {
      "epoch": 7.028765481422293,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 17593
    },
    {
      "epoch": 7.029165001997603,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 17594
    },
    {
      "epoch": 7.029564522572913,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 17595
    },
    {
      "epoch": 7.029964043148222,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 17596
    },
    {
      "epoch": 7.030363563723531,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 17597
    },
    {
      "epoch": 7.030763084298841,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 17598
    },
    {
      "epoch": 7.031162604874151,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 17599
    },
    {
      "epoch": 7.031562125449461,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 17600
    },
    {
      "epoch": 7.0319616460247705,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 17601
    },
    {
      "epoch": 7.03236116660008,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 17602
    },
    {
      "epoch": 7.032760687175389,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 17603
    },
    {
      "epoch": 7.033160207750699,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 17604
    },
    {
      "epoch": 7.033559728326009,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 17605
    },
    {
      "epoch": 7.0339592489013185,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 17606
    },
    {
      "epoch": 7.034358769476628,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 17607
    },
    {
      "epoch": 7.034758290051938,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 17608
    },
    {
      "epoch": 7.035157810627247,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 17609
    },
    {
      "epoch": 7.035557331202557,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 17610
    },
    {
      "epoch": 7.035956851777867,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 17611
    },
    {
      "epoch": 7.036356372353176,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 17612
    },
    {
      "epoch": 7.036755892928486,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 17613
    },
    {
      "epoch": 7.037155413503795,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 17614
    },
    {
      "epoch": 7.037554934079105,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 17615
    },
    {
      "epoch": 7.037954454654415,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 17616
    },
    {
      "epoch": 7.038353975229724,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 17617
    },
    {
      "epoch": 7.038753495805034,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 17618
    },
    {
      "epoch": 7.039153016380344,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 17619
    },
    {
      "epoch": 7.039552536955653,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 17620
    },
    {
      "epoch": 7.039952057530963,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 17621
    },
    {
      "epoch": 7.040351578106272,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 17622
    },
    {
      "epoch": 7.040751098681582,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 17623
    },
    {
      "epoch": 7.041150619256892,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 17624
    },
    {
      "epoch": 7.041550139832202,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 17625
    },
    {
      "epoch": 7.041949660407511,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 17626
    },
    {
      "epoch": 7.0423491809828205,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 17627
    },
    {
      "epoch": 7.04274870155813,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 17628
    },
    {
      "epoch": 7.04314822213344,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 17629
    },
    {
      "epoch": 7.04354774270875,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 17630
    },
    {
      "epoch": 7.043947263284059,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 17631
    },
    {
      "epoch": 7.0443467838593685,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 17632
    },
    {
      "epoch": 7.044746304434678,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 17633
    },
    {
      "epoch": 7.045145825009988,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 17634
    },
    {
      "epoch": 7.045545345585298,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 17635
    },
    {
      "epoch": 7.045944866160608,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 17636
    },
    {
      "epoch": 7.046344386735917,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 17637
    },
    {
      "epoch": 7.046743907311226,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 17638
    },
    {
      "epoch": 7.047143427886536,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 17639
    },
    {
      "epoch": 7.047542948461846,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 17640
    },
    {
      "epoch": 7.047942469037156,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 17641
    },
    {
      "epoch": 7.048341989612465,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 17642
    },
    {
      "epoch": 7.048741510187774,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 17643
    },
    {
      "epoch": 7.049141030763084,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 17644
    },
    {
      "epoch": 7.049540551338394,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 17645
    },
    {
      "epoch": 7.049940071913704,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 17646
    },
    {
      "epoch": 7.0503395924890135,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 17647
    },
    {
      "epoch": 7.050739113064322,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 17648
    },
    {
      "epoch": 7.051138633639632,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 17649
    },
    {
      "epoch": 7.051538154214942,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 17650
    },
    {
      "epoch": 7.051937674790252,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 17651
    },
    {
      "epoch": 7.052337195365562,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 17652
    },
    {
      "epoch": 7.052736715940871,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 17653
    },
    {
      "epoch": 7.05313623651618,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 17654
    },
    {
      "epoch": 7.05353575709149,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 17655
    },
    {
      "epoch": 7.0539352776668,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 17656
    },
    {
      "epoch": 7.05433479824211,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 17657
    },
    {
      "epoch": 7.054734318817419,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 17658
    },
    {
      "epoch": 7.055133839392728,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 17659
    },
    {
      "epoch": 7.055533359968038,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 17660
    },
    {
      "epoch": 7.055932880543348,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 17661
    },
    {
      "epoch": 7.056332401118658,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 17662
    },
    {
      "epoch": 7.0567319216939675,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 17663
    },
    {
      "epoch": 7.057131442269277,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 17664
    },
    {
      "epoch": 7.057530962844586,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 17665
    },
    {
      "epoch": 7.057930483419896,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 17666
    },
    {
      "epoch": 7.058330003995206,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 17667
    },
    {
      "epoch": 7.0587295245705155,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 17668
    },
    {
      "epoch": 7.059129045145825,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 17669
    },
    {
      "epoch": 7.059528565721135,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 17670
    },
    {
      "epoch": 7.059928086296444,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 17671
    },
    {
      "epoch": 7.060327606871754,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 17672
    },
    {
      "epoch": 7.0607271274470635,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 17673
    },
    {
      "epoch": 7.061126648022373,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 17674
    },
    {
      "epoch": 7.061526168597683,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 17675
    },
    {
      "epoch": 7.061925689172992,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 17676
    },
    {
      "epoch": 7.062325209748302,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 17677
    },
    {
      "epoch": 7.062724730323612,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 17678
    },
    {
      "epoch": 7.063124250898921,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 17679
    },
    {
      "epoch": 7.063523771474231,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 17680
    },
    {
      "epoch": 7.063923292049541,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 17681
    },
    {
      "epoch": 7.06432281262485,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 17682
    },
    {
      "epoch": 7.06472233320016,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 17683
    },
    {
      "epoch": 7.065121853775469,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 17684
    },
    {
      "epoch": 7.065521374350779,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 17685
    },
    {
      "epoch": 7.065920894926089,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 17686
    },
    {
      "epoch": 7.066320415501398,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 17687
    },
    {
      "epoch": 7.066719936076708,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 17688
    },
    {
      "epoch": 7.0671194566520175,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 17689
    },
    {
      "epoch": 7.067518977227327,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 17690
    },
    {
      "epoch": 7.067918497802637,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 17691
    },
    {
      "epoch": 7.068318018377947,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 17692
    },
    {
      "epoch": 7.068717538953256,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 17693
    },
    {
      "epoch": 7.0691170595285655,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 17694
    },
    {
      "epoch": 7.069516580103875,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 17695
    },
    {
      "epoch": 7.069916100679185,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 17696
    },
    {
      "epoch": 7.070315621254495,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 17697
    },
    {
      "epoch": 7.070715141829805,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 17698
    },
    {
      "epoch": 7.0711146624051135,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.054,
      "step": 17699
    },
    {
      "epoch": 7.071514182980423,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 17700
    },
    {
      "epoch": 7.071913703555733,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 17701
    },
    {
      "epoch": 7.072313224131043,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 17702
    },
    {
      "epoch": 7.072712744706353,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 17703
    },
    {
      "epoch": 7.073112265281662,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 17704
    },
    {
      "epoch": 7.073511785856971,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 17705
    },
    {
      "epoch": 7.073911306432281,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 17706
    },
    {
      "epoch": 7.074310827007591,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 17707
    },
    {
      "epoch": 7.074710347582901,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 17708
    },
    {
      "epoch": 7.0751098681582105,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 17709
    },
    {
      "epoch": 7.075509388733519,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 17710
    },
    {
      "epoch": 7.075908909308829,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 17711
    },
    {
      "epoch": 7.076308429884139,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 17712
    },
    {
      "epoch": 7.076707950459449,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 17713
    },
    {
      "epoch": 7.077107471034759,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 17714
    },
    {
      "epoch": 7.077506991610068,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 17715
    },
    {
      "epoch": 7.077906512185377,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 17716
    },
    {
      "epoch": 7.078306032760687,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 17717
    },
    {
      "epoch": 7.078705553335997,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 17718
    },
    {
      "epoch": 7.079105073911307,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 17719
    },
    {
      "epoch": 7.079504594486616,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 17720
    },
    {
      "epoch": 7.079904115061925,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 17721
    },
    {
      "epoch": 7.080303635637235,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 17722
    },
    {
      "epoch": 7.080703156212545,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 17723
    },
    {
      "epoch": 7.081102676787855,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 17724
    },
    {
      "epoch": 7.081502197363164,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 17725
    },
    {
      "epoch": 7.081901717938474,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 17726
    },
    {
      "epoch": 7.082301238513783,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 17727
    },
    {
      "epoch": 7.082700759089093,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 17728
    },
    {
      "epoch": 7.083100279664403,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 17729
    },
    {
      "epoch": 7.0834998002397125,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 17730
    },
    {
      "epoch": 7.083899320815022,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 17731
    },
    {
      "epoch": 7.084298841390331,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 17732
    },
    {
      "epoch": 7.084698361965641,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 17733
    },
    {
      "epoch": 7.085097882540951,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 17734
    },
    {
      "epoch": 7.0854974031162605,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 17735
    },
    {
      "epoch": 7.08589692369157,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 17736
    },
    {
      "epoch": 7.08629644426688,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 17737
    },
    {
      "epoch": 7.086695964842189,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 17738
    },
    {
      "epoch": 7.087095485417499,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 17739
    },
    {
      "epoch": 7.087495005992809,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 17740
    },
    {
      "epoch": 7.087894526568118,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 17741
    },
    {
      "epoch": 7.088294047143428,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 17742
    },
    {
      "epoch": 7.088693567718738,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 17743
    },
    {
      "epoch": 7.089093088294047,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 17744
    },
    {
      "epoch": 7.089492608869357,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 17745
    },
    {
      "epoch": 7.089892129444666,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 17746
    },
    {
      "epoch": 7.090291650019976,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 17747
    },
    {
      "epoch": 7.090691170595286,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 17748
    },
    {
      "epoch": 7.091090691170595,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 17749
    },
    {
      "epoch": 7.091490211745905,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 17750
    },
    {
      "epoch": 7.0918897323212144,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 17751
    },
    {
      "epoch": 7.092289252896524,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 17752
    },
    {
      "epoch": 7.092688773471834,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 17753
    },
    {
      "epoch": 7.093088294047144,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 17754
    },
    {
      "epoch": 7.093487814622453,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 17755
    },
    {
      "epoch": 7.0938873351977625,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 17756
    },
    {
      "epoch": 7.094286855773072,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 17757
    },
    {
      "epoch": 7.094686376348382,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 17758
    },
    {
      "epoch": 7.095085896923692,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 17759
    },
    {
      "epoch": 7.095485417499002,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 17760
    },
    {
      "epoch": 7.0958849380743105,
      "grad_norm": 0.171875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 17761
    },
    {
      "epoch": 7.09628445864962,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 17762
    },
    {
      "epoch": 7.09668397922493,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 17763
    },
    {
      "epoch": 7.09708349980024,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 17764
    },
    {
      "epoch": 7.09748302037555,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 17765
    },
    {
      "epoch": 7.097882540950859,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 17766
    },
    {
      "epoch": 7.098282061526168,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 17767
    },
    {
      "epoch": 7.098681582101478,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 17768
    },
    {
      "epoch": 7.099081102676788,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 17769
    },
    {
      "epoch": 7.099480623252098,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 17770
    },
    {
      "epoch": 7.0998801438274075,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 17771
    },
    {
      "epoch": 7.100279664402716,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 17772
    },
    {
      "epoch": 7.100679184978026,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 17773
    },
    {
      "epoch": 7.101078705553336,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 17774
    },
    {
      "epoch": 7.101478226128646,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 17775
    },
    {
      "epoch": 7.1018777467039556,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 17776
    },
    {
      "epoch": 7.1022772672792644,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 17777
    },
    {
      "epoch": 7.102676787854574,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 17778
    },
    {
      "epoch": 7.103076308429884,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 17779
    },
    {
      "epoch": 7.103475829005194,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 17780
    },
    {
      "epoch": 7.103875349580504,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 17781
    },
    {
      "epoch": 7.104274870155813,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 17782
    },
    {
      "epoch": 7.104674390731122,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 17783
    },
    {
      "epoch": 7.105073911306432,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 17784
    },
    {
      "epoch": 7.105473431881742,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 17785
    },
    {
      "epoch": 7.105872952457052,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 17786
    },
    {
      "epoch": 7.106272473032361,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 17787
    },
    {
      "epoch": 7.106671993607671,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 17788
    },
    {
      "epoch": 7.10707151418298,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 17789
    },
    {
      "epoch": 7.10747103475829,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 17790
    },
    {
      "epoch": 7.1078705553336,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 17791
    },
    {
      "epoch": 7.1082700759089095,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 17792
    },
    {
      "epoch": 7.108669596484219,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 17793
    },
    {
      "epoch": 7.109069117059528,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 17794
    },
    {
      "epoch": 7.109468637634838,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 17795
    },
    {
      "epoch": 7.109868158210148,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 17796
    },
    {
      "epoch": 7.1102676787854575,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 17797
    },
    {
      "epoch": 7.110667199360767,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 17798
    },
    {
      "epoch": 7.111066719936077,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 17799
    },
    {
      "epoch": 7.111466240511386,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 17800
    },
    {
      "epoch": 7.111865761086696,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 17801
    },
    {
      "epoch": 7.112265281662006,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 17802
    },
    {
      "epoch": 7.112664802237315,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 17803
    },
    {
      "epoch": 7.113064322812625,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 17804
    },
    {
      "epoch": 7.113463843387935,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 17805
    },
    {
      "epoch": 7.113863363963244,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 17806
    },
    {
      "epoch": 7.114262884538554,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 17807
    },
    {
      "epoch": 7.114662405113863,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 17808
    },
    {
      "epoch": 7.115061925689173,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 17809
    },
    {
      "epoch": 7.115461446264483,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 17810
    },
    {
      "epoch": 7.115860966839792,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 17811
    },
    {
      "epoch": 7.116260487415102,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 17812
    },
    {
      "epoch": 7.116660007990411,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 17813
    },
    {
      "epoch": 7.117059528565721,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 17814
    },
    {
      "epoch": 7.117459049141031,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 17815
    },
    {
      "epoch": 7.117858569716341,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 17816
    },
    {
      "epoch": 7.11825809029165,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 17817
    },
    {
      "epoch": 7.1186576108669595,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 17818
    },
    {
      "epoch": 7.119057131442269,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 17819
    },
    {
      "epoch": 7.119456652017579,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 17820
    },
    {
      "epoch": 7.119856172592889,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 17821
    },
    {
      "epoch": 7.120255693168198,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 17822
    },
    {
      "epoch": 7.1206552137435075,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 17823
    },
    {
      "epoch": 7.121054734318817,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 17824
    },
    {
      "epoch": 7.121454254894127,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0565,
      "step": 17825
    },
    {
      "epoch": 7.121853775469437,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 17826
    },
    {
      "epoch": 7.122253296044747,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 17827
    },
    {
      "epoch": 7.122652816620056,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 17828
    },
    {
      "epoch": 7.123052337195365,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 17829
    },
    {
      "epoch": 7.123451857770675,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 17830
    },
    {
      "epoch": 7.123851378345985,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 17831
    },
    {
      "epoch": 7.124250898921295,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 17832
    },
    {
      "epoch": 7.1246504194966045,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 17833
    },
    {
      "epoch": 7.125049940071913,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 17834
    },
    {
      "epoch": 7.125449460647223,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 17835
    },
    {
      "epoch": 7.125848981222533,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 17836
    },
    {
      "epoch": 7.126248501797843,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 17837
    },
    {
      "epoch": 7.1266480223731525,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 17838
    },
    {
      "epoch": 7.127047542948461,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 17839
    },
    {
      "epoch": 7.127447063523771,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 17840
    },
    {
      "epoch": 7.127846584099081,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 17841
    },
    {
      "epoch": 7.128246104674391,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 17842
    },
    {
      "epoch": 7.128645625249701,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 17843
    },
    {
      "epoch": 7.12904514582501,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 17844
    },
    {
      "epoch": 7.129444666400319,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 17845
    },
    {
      "epoch": 7.129844186975629,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 17846
    },
    {
      "epoch": 7.130243707550939,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 17847
    },
    {
      "epoch": 7.130643228126249,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 17848
    },
    {
      "epoch": 7.131042748701558,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 17849
    },
    {
      "epoch": 7.131442269276867,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 17850
    },
    {
      "epoch": 7.131841789852177,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 17851
    },
    {
      "epoch": 7.132241310427487,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 17852
    },
    {
      "epoch": 7.132640831002797,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 17853
    },
    {
      "epoch": 7.1330403515781065,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 17854
    },
    {
      "epoch": 7.133439872153416,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 17855
    },
    {
      "epoch": 7.133839392728725,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 17856
    },
    {
      "epoch": 7.134238913304035,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 17857
    },
    {
      "epoch": 7.134638433879345,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 17858
    },
    {
      "epoch": 7.1350379544546545,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 17859
    },
    {
      "epoch": 7.135437475029964,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 17860
    },
    {
      "epoch": 7.135836995605274,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 17861
    },
    {
      "epoch": 7.136236516180583,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 17862
    },
    {
      "epoch": 7.136636036755893,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 17863
    },
    {
      "epoch": 7.1370355573312025,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 17864
    },
    {
      "epoch": 7.137435077906512,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 17865
    },
    {
      "epoch": 7.137834598481822,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 17866
    },
    {
      "epoch": 7.138234119057131,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 17867
    },
    {
      "epoch": 7.138633639632441,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 17868
    },
    {
      "epoch": 7.139033160207751,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 17869
    },
    {
      "epoch": 7.13943268078306,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 17870
    },
    {
      "epoch": 7.13983220135837,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 17871
    },
    {
      "epoch": 7.14023172193368,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 17872
    },
    {
      "epoch": 7.140631242508989,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 17873
    },
    {
      "epoch": 7.141030763084299,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 17874
    },
    {
      "epoch": 7.141430283659608,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 17875
    },
    {
      "epoch": 7.141829804234918,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 17876
    },
    {
      "epoch": 7.142229324810228,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 17877
    },
    {
      "epoch": 7.142628845385538,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 17878
    },
    {
      "epoch": 7.143028365960847,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 17879
    },
    {
      "epoch": 7.1434278865361565,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 17880
    },
    {
      "epoch": 7.143827407111466,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 17881
    },
    {
      "epoch": 7.144226927686776,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 17882
    },
    {
      "epoch": 7.144626448262086,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 17883
    },
    {
      "epoch": 7.145025968837395,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 17884
    },
    {
      "epoch": 7.1454254894127045,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 17885
    },
    {
      "epoch": 7.145825009988014,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 17886
    },
    {
      "epoch": 7.146224530563324,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 17887
    },
    {
      "epoch": 7.146624051138634,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 17888
    },
    {
      "epoch": 7.147023571713944,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 17889
    },
    {
      "epoch": 7.1474230922892525,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 17890
    },
    {
      "epoch": 7.147822612864562,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 17891
    },
    {
      "epoch": 7.148222133439872,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 17892
    },
    {
      "epoch": 7.148621654015182,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 17893
    },
    {
      "epoch": 7.149021174590492,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 17894
    },
    {
      "epoch": 7.1494206951658015,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 17895
    },
    {
      "epoch": 7.14982021574111,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 17896
    },
    {
      "epoch": 7.15021973631642,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 17897
    },
    {
      "epoch": 7.15061925689173,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 17898
    },
    {
      "epoch": 7.15101877746704,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 17899
    },
    {
      "epoch": 7.1514182980423495,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 17900
    },
    {
      "epoch": 7.151817818617658,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 17901
    },
    {
      "epoch": 7.152217339192968,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 17902
    },
    {
      "epoch": 7.152616859768278,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 17903
    },
    {
      "epoch": 7.153016380343588,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0567,
      "step": 17904
    },
    {
      "epoch": 7.153415900918898,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 17905
    },
    {
      "epoch": 7.153815421494207,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 17906
    },
    {
      "epoch": 7.154214942069516,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 17907
    },
    {
      "epoch": 7.154614462644826,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 17908
    },
    {
      "epoch": 7.155013983220136,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 17909
    },
    {
      "epoch": 7.155413503795446,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 17910
    },
    {
      "epoch": 7.155813024370755,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 17911
    },
    {
      "epoch": 7.156212544946065,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 17912
    },
    {
      "epoch": 7.156612065521374,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 17913
    },
    {
      "epoch": 7.157011586096684,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 17914
    },
    {
      "epoch": 7.157411106671994,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 17915
    },
    {
      "epoch": 7.157810627247303,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 17916
    },
    {
      "epoch": 7.158210147822613,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 17917
    },
    {
      "epoch": 7.158609668397922,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 17918
    },
    {
      "epoch": 7.159009188973232,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 17919
    },
    {
      "epoch": 7.159408709548542,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 17920
    },
    {
      "epoch": 7.1598082301238515,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 17921
    },
    {
      "epoch": 7.160207750699161,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 17922
    },
    {
      "epoch": 7.160607271274471,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 17923
    },
    {
      "epoch": 7.16100679184978,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 17924
    },
    {
      "epoch": 7.16140631242509,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 17925
    },
    {
      "epoch": 7.1618058330003995,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 17926
    },
    {
      "epoch": 7.162205353575709,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 17927
    },
    {
      "epoch": 7.162604874151019,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 17928
    },
    {
      "epoch": 7.163004394726328,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 17929
    },
    {
      "epoch": 7.163403915301638,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 17930
    },
    {
      "epoch": 7.163803435876948,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 17931
    },
    {
      "epoch": 7.164202956452257,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 17932
    },
    {
      "epoch": 7.164602477027567,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 17933
    },
    {
      "epoch": 7.165001997602877,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 17934
    },
    {
      "epoch": 7.165401518178186,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 17935
    },
    {
      "epoch": 7.165801038753496,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 17936
    },
    {
      "epoch": 7.166200559328805,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 17937
    },
    {
      "epoch": 7.166600079904115,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 17938
    },
    {
      "epoch": 7.166999600479425,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 17939
    },
    {
      "epoch": 7.167399121054734,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 17940
    },
    {
      "epoch": 7.167798641630044,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 17941
    },
    {
      "epoch": 7.168198162205353,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 17942
    },
    {
      "epoch": 7.168597682780663,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 17943
    },
    {
      "epoch": 7.168997203355973,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 17944
    },
    {
      "epoch": 7.169396723931283,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0577,
      "step": 17945
    },
    {
      "epoch": 7.169796244506592,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.059,
      "step": 17946
    },
    {
      "epoch": 7.1701957650819015,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 17947
    },
    {
      "epoch": 7.170595285657211,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 17948
    },
    {
      "epoch": 7.170994806232521,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 17949
    },
    {
      "epoch": 7.171394326807831,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 17950
    },
    {
      "epoch": 7.171793847383141,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 17951
    },
    {
      "epoch": 7.1721933679584495,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 17952
    },
    {
      "epoch": 7.172592888533759,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 17953
    },
    {
      "epoch": 7.172992409109069,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 17954
    },
    {
      "epoch": 7.173391929684379,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 17955
    },
    {
      "epoch": 7.173791450259689,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 17956
    },
    {
      "epoch": 7.174190970834998,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 17957
    },
    {
      "epoch": 7.174590491410307,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 17958
    },
    {
      "epoch": 7.174990011985617,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 17959
    },
    {
      "epoch": 7.175389532560927,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 17960
    },
    {
      "epoch": 7.175789053136237,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 17961
    },
    {
      "epoch": 7.1761885737115465,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 17962
    },
    {
      "epoch": 7.176588094286855,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 17963
    },
    {
      "epoch": 7.176987614862165,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 17964
    },
    {
      "epoch": 7.177387135437475,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 17965
    },
    {
      "epoch": 7.177786656012785,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 17966
    },
    {
      "epoch": 7.1781861765880945,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 17967
    },
    {
      "epoch": 7.178585697163404,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 17968
    },
    {
      "epoch": 7.178985217738713,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 17969
    },
    {
      "epoch": 7.179384738314023,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 17970
    },
    {
      "epoch": 7.179784258889333,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 17971
    },
    {
      "epoch": 7.180183779464643,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 17972
    },
    {
      "epoch": 7.180583300039952,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 17973
    },
    {
      "epoch": 7.180982820615261,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 17974
    },
    {
      "epoch": 7.181382341190571,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 17975
    },
    {
      "epoch": 7.181781861765881,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 17976
    },
    {
      "epoch": 7.182181382341191,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 17977
    },
    {
      "epoch": 7.1825809029165,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 17978
    },
    {
      "epoch": 7.18298042349181,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 17979
    },
    {
      "epoch": 7.183379944067119,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 17980
    },
    {
      "epoch": 7.183779464642429,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 17981
    },
    {
      "epoch": 7.184178985217739,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 17982
    },
    {
      "epoch": 7.1845785057930485,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 17983
    },
    {
      "epoch": 7.184978026368358,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 17984
    },
    {
      "epoch": 7.185377546943668,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 17985
    },
    {
      "epoch": 7.185777067518977,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 17986
    },
    {
      "epoch": 7.186176588094287,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 17987
    },
    {
      "epoch": 7.1865761086695965,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 17988
    },
    {
      "epoch": 7.186975629244906,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 17989
    },
    {
      "epoch": 7.187375149820216,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 17990
    },
    {
      "epoch": 7.187774670395525,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 17991
    },
    {
      "epoch": 7.188174190970835,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 17992
    },
    {
      "epoch": 7.1885737115461446,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 17993
    },
    {
      "epoch": 7.188973232121454,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 17994
    },
    {
      "epoch": 7.189372752696764,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 17995
    },
    {
      "epoch": 7.189772273272074,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 17996
    },
    {
      "epoch": 7.190171793847383,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 17997
    },
    {
      "epoch": 7.190571314422693,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 17998
    },
    {
      "epoch": 7.190970834998002,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 17999
    },
    {
      "epoch": 7.191370355573312,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 18000
    },
    {
      "epoch": 7.191769876148622,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 18001
    },
    {
      "epoch": 7.192169396723932,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 18002
    },
    {
      "epoch": 7.192568917299241,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 18003
    },
    {
      "epoch": 7.19296843787455,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 18004
    },
    {
      "epoch": 7.19336795844986,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 18005
    },
    {
      "epoch": 7.19376747902517,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 18006
    },
    {
      "epoch": 7.19416699960048,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 18007
    },
    {
      "epoch": 7.194566520175789,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 18008
    },
    {
      "epoch": 7.1949660407510985,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 18009
    },
    {
      "epoch": 7.195365561326408,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 18010
    },
    {
      "epoch": 7.195765081901718,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 18011
    },
    {
      "epoch": 7.196164602477028,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 18012
    },
    {
      "epoch": 7.196564123052338,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 18013
    },
    {
      "epoch": 7.1969636436276465,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 18014
    },
    {
      "epoch": 7.197363164202956,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 18015
    },
    {
      "epoch": 7.197762684778266,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 18016
    },
    {
      "epoch": 7.198162205353576,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 18017
    },
    {
      "epoch": 7.198561725928886,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 18018
    },
    {
      "epoch": 7.1989612465041946,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 18019
    },
    {
      "epoch": 7.199360767079504,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 18020
    },
    {
      "epoch": 7.199760287654814,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 18021
    },
    {
      "epoch": 7.200159808230124,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 18022
    },
    {
      "epoch": 7.200559328805434,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 18023
    },
    {
      "epoch": 7.2009588493807435,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 18024
    },
    {
      "epoch": 7.201358369956052,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 18025
    },
    {
      "epoch": 7.201757890531362,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 18026
    },
    {
      "epoch": 7.202157411106672,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 18027
    },
    {
      "epoch": 7.202556931681982,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 18028
    },
    {
      "epoch": 7.2029564522572915,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 18029
    },
    {
      "epoch": 7.2033559728326,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 18030
    },
    {
      "epoch": 7.20375549340791,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 18031
    },
    {
      "epoch": 7.20415501398322,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 18032
    },
    {
      "epoch": 7.20455453455853,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 18033
    },
    {
      "epoch": 7.20495405513384,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 18034
    },
    {
      "epoch": 7.205353575709149,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 18035
    },
    {
      "epoch": 7.205753096284458,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 18036
    },
    {
      "epoch": 7.206152616859768,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 18037
    },
    {
      "epoch": 7.206552137435078,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 18038
    },
    {
      "epoch": 7.206951658010388,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 18039
    },
    {
      "epoch": 7.207351178585697,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 18040
    },
    {
      "epoch": 7.207750699161007,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 18041
    },
    {
      "epoch": 7.208150219736316,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 18042
    },
    {
      "epoch": 7.208549740311626,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 18043
    },
    {
      "epoch": 7.208949260886936,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 18044
    },
    {
      "epoch": 7.2093487814622454,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 18045
    },
    {
      "epoch": 7.209748302037555,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 18046
    },
    {
      "epoch": 7.210147822612864,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 18047
    },
    {
      "epoch": 7.210547343188174,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 18048
    },
    {
      "epoch": 7.210946863763484,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 18049
    },
    {
      "epoch": 7.2113463843387935,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 18050
    },
    {
      "epoch": 7.211745904914103,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 18051
    },
    {
      "epoch": 7.212145425489413,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 18052
    },
    {
      "epoch": 7.212544946064722,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 18053
    },
    {
      "epoch": 7.212944466640032,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 18054
    },
    {
      "epoch": 7.2133439872153415,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 18055
    },
    {
      "epoch": 7.213743507790651,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 18056
    },
    {
      "epoch": 7.214143028365961,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 18057
    },
    {
      "epoch": 7.214542548941271,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 18058
    },
    {
      "epoch": 7.21494206951658,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 18059
    },
    {
      "epoch": 7.21534159009189,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 18060
    },
    {
      "epoch": 7.215741110667199,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 18061
    },
    {
      "epoch": 7.216140631242509,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 18062
    },
    {
      "epoch": 7.216540151817819,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 18063
    },
    {
      "epoch": 7.216939672393128,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 18064
    },
    {
      "epoch": 7.217339192968438,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 18065
    },
    {
      "epoch": 7.217738713543747,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 18066
    },
    {
      "epoch": 7.218138234119057,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 18067
    },
    {
      "epoch": 7.218537754694367,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 18068
    },
    {
      "epoch": 7.218937275269677,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 18069
    },
    {
      "epoch": 7.219336795844986,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 18070
    },
    {
      "epoch": 7.2197363164202955,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 18071
    },
    {
      "epoch": 7.220135836995605,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 18072
    },
    {
      "epoch": 7.220535357570915,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 18073
    },
    {
      "epoch": 7.220934878146225,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 18074
    },
    {
      "epoch": 7.221334398721535,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 18075
    },
    {
      "epoch": 7.2217339192968435,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 18076
    },
    {
      "epoch": 7.222133439872153,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 18077
    },
    {
      "epoch": 7.222532960447463,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 18078
    },
    {
      "epoch": 7.222932481022773,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 18079
    },
    {
      "epoch": 7.223332001598083,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 18080
    },
    {
      "epoch": 7.2237315221733915,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 18081
    },
    {
      "epoch": 7.224131042748701,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 18082
    },
    {
      "epoch": 7.224530563324011,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 18083
    },
    {
      "epoch": 7.224930083899321,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 18084
    },
    {
      "epoch": 7.225329604474631,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 18085
    },
    {
      "epoch": 7.2257291250499405,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 18086
    },
    {
      "epoch": 7.226128645625249,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 18087
    },
    {
      "epoch": 7.226528166200559,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 18088
    },
    {
      "epoch": 7.226927686775869,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 18089
    },
    {
      "epoch": 7.227327207351179,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 18090
    },
    {
      "epoch": 7.2277267279264885,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 18091
    },
    {
      "epoch": 7.228126248501797,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 18092
    },
    {
      "epoch": 7.228525769077107,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 18093
    },
    {
      "epoch": 7.228925289652417,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 18094
    },
    {
      "epoch": 7.229324810227727,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 18095
    },
    {
      "epoch": 7.229724330803037,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 18096
    },
    {
      "epoch": 7.230123851378346,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 18097
    },
    {
      "epoch": 7.230523371953655,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 18098
    },
    {
      "epoch": 7.230922892528965,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 18099
    },
    {
      "epoch": 7.231322413104275,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 18100
    },
    {
      "epoch": 7.231721933679585,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 18101
    },
    {
      "epoch": 7.232121454254894,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 18102
    },
    {
      "epoch": 7.232520974830204,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 18103
    },
    {
      "epoch": 7.232920495405513,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 18104
    },
    {
      "epoch": 7.233320015980823,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 18105
    },
    {
      "epoch": 7.233719536556133,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 18106
    },
    {
      "epoch": 7.234119057131442,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 18107
    },
    {
      "epoch": 7.234518577706752,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 18108
    },
    {
      "epoch": 7.234918098282061,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 18109
    },
    {
      "epoch": 7.235317618857371,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 18110
    },
    {
      "epoch": 7.235717139432681,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 18111
    },
    {
      "epoch": 7.2361166600079905,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 18112
    },
    {
      "epoch": 7.2365161805833,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 18113
    },
    {
      "epoch": 7.23691570115861,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 18114
    },
    {
      "epoch": 7.237315221733919,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 18115
    },
    {
      "epoch": 7.237714742309229,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 18116
    },
    {
      "epoch": 7.2381142628845385,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 18117
    },
    {
      "epoch": 7.238513783459848,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 18118
    },
    {
      "epoch": 7.238913304035158,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 18119
    },
    {
      "epoch": 7.239312824610467,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 18120
    },
    {
      "epoch": 7.239712345185777,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 18121
    },
    {
      "epoch": 7.240111865761087,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 18122
    },
    {
      "epoch": 7.240511386336396,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 18123
    },
    {
      "epoch": 7.240910906911706,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 18124
    },
    {
      "epoch": 7.241310427487016,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 18125
    },
    {
      "epoch": 7.241709948062325,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 18126
    },
    {
      "epoch": 7.242109468637635,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 18127
    },
    {
      "epoch": 7.242508989212944,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 18128
    },
    {
      "epoch": 7.242908509788254,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 18129
    },
    {
      "epoch": 7.243308030363564,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0564,
      "step": 18130
    },
    {
      "epoch": 7.243707550938874,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 18131
    },
    {
      "epoch": 7.244107071514183,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 18132
    },
    {
      "epoch": 7.244506592089492,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 18133
    },
    {
      "epoch": 7.244906112664802,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 18134
    },
    {
      "epoch": 7.245305633240112,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 18135
    },
    {
      "epoch": 7.245705153815422,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 18136
    },
    {
      "epoch": 7.246104674390731,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 18137
    },
    {
      "epoch": 7.2465041949660405,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 18138
    },
    {
      "epoch": 7.24690371554135,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 18139
    },
    {
      "epoch": 7.24730323611666,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 18140
    },
    {
      "epoch": 7.24770275669197,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 18141
    },
    {
      "epoch": 7.24810227726728,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 18142
    },
    {
      "epoch": 7.2485017978425885,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 18143
    },
    {
      "epoch": 7.248901318417898,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 18144
    },
    {
      "epoch": 7.249300838993208,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 18145
    },
    {
      "epoch": 7.249700359568518,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 18146
    },
    {
      "epoch": 7.250099880143828,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 18147
    },
    {
      "epoch": 7.2504994007191375,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 18148
    },
    {
      "epoch": 7.250898921294446,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 18149
    },
    {
      "epoch": 7.251298441869756,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 18150
    },
    {
      "epoch": 7.251697962445066,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 18151
    },
    {
      "epoch": 7.252097483020376,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 18152
    },
    {
      "epoch": 7.2524970035956855,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 18153
    },
    {
      "epoch": 7.252896524170994,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 18154
    },
    {
      "epoch": 7.253296044746304,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 18155
    },
    {
      "epoch": 7.253695565321614,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 18156
    },
    {
      "epoch": 7.254095085896924,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 18157
    },
    {
      "epoch": 7.2544946064722335,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 18158
    },
    {
      "epoch": 7.254894127047543,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 18159
    },
    {
      "epoch": 7.255293647622852,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 18160
    },
    {
      "epoch": 7.255693168198162,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 18161
    },
    {
      "epoch": 7.256092688773472,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 18162
    },
    {
      "epoch": 7.256492209348782,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 18163
    },
    {
      "epoch": 7.256891729924091,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 18164
    },
    {
      "epoch": 7.257291250499401,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 18165
    },
    {
      "epoch": 7.25769077107471,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 18166
    },
    {
      "epoch": 7.25809029165002,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 18167
    },
    {
      "epoch": 7.25848981222533,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 18168
    },
    {
      "epoch": 7.258889332800639,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 18169
    },
    {
      "epoch": 7.259288853375949,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 18170
    },
    {
      "epoch": 7.259688373951258,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 18171
    },
    {
      "epoch": 7.260087894526568,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 18172
    },
    {
      "epoch": 7.260487415101878,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 18173
    },
    {
      "epoch": 7.2608869356771875,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 18174
    },
    {
      "epoch": 7.261286456252497,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 18175
    },
    {
      "epoch": 7.261685976827807,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 18176
    },
    {
      "epoch": 7.262085497403116,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 18177
    },
    {
      "epoch": 7.262485017978426,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 18178
    },
    {
      "epoch": 7.2628845385537355,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 18179
    },
    {
      "epoch": 7.263284059129045,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 18180
    },
    {
      "epoch": 7.263683579704355,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 18181
    },
    {
      "epoch": 7.264083100279665,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 18182
    },
    {
      "epoch": 7.264482620854974,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 18183
    },
    {
      "epoch": 7.2648821414302835,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 18184
    },
    {
      "epoch": 7.265281662005593,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 18185
    },
    {
      "epoch": 7.265681182580903,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 18186
    },
    {
      "epoch": 7.266080703156213,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 18187
    },
    {
      "epoch": 7.266480223731522,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 18188
    },
    {
      "epoch": 7.266879744306832,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 18189
    },
    {
      "epoch": 7.267279264882141,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 18190
    },
    {
      "epoch": 7.267678785457451,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 18191
    },
    {
      "epoch": 7.268078306032761,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 18192
    },
    {
      "epoch": 7.26847782660807,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 18193
    },
    {
      "epoch": 7.26887734718338,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 18194
    },
    {
      "epoch": 7.269276867758689,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 18195
    },
    {
      "epoch": 7.269676388333999,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 18196
    },
    {
      "epoch": 7.270075908909309,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 18197
    },
    {
      "epoch": 7.270475429484619,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 18198
    },
    {
      "epoch": 7.270874950059928,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 18199
    },
    {
      "epoch": 7.2712744706352375,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 18200
    },
    {
      "epoch": 7.271673991210547,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 18201
    },
    {
      "epoch": 7.272073511785857,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 18202
    },
    {
      "epoch": 7.272473032361167,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 18203
    },
    {
      "epoch": 7.272872552936477,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 18204
    },
    {
      "epoch": 7.2732720735117855,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 18205
    },
    {
      "epoch": 7.273671594087095,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 18206
    },
    {
      "epoch": 7.274071114662405,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 18207
    },
    {
      "epoch": 7.274470635237715,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 18208
    },
    {
      "epoch": 7.274870155813025,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 18209
    },
    {
      "epoch": 7.2752696763883336,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 18210
    },
    {
      "epoch": 7.275669196963643,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 18211
    },
    {
      "epoch": 7.276068717538953,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 18212
    },
    {
      "epoch": 7.276468238114263,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 18213
    },
    {
      "epoch": 7.276867758689573,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 18214
    },
    {
      "epoch": 7.2772672792648825,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 18215
    },
    {
      "epoch": 7.277666799840191,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 18216
    },
    {
      "epoch": 7.278066320415501,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 18217
    },
    {
      "epoch": 7.278465840990811,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 18218
    },
    {
      "epoch": 7.278865361566121,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 18219
    },
    {
      "epoch": 7.2792648821414305,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 18220
    },
    {
      "epoch": 7.27966440271674,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 18221
    },
    {
      "epoch": 7.280063923292049,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0594,
      "step": 18222
    },
    {
      "epoch": 7.280463443867359,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 18223
    },
    {
      "epoch": 7.280862964442669,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 18224
    },
    {
      "epoch": 7.281262485017979,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 18225
    },
    {
      "epoch": 7.281662005593288,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 18226
    },
    {
      "epoch": 7.282061526168597,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 18227
    },
    {
      "epoch": 7.282461046743907,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 18228
    },
    {
      "epoch": 7.282860567319217,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 18229
    },
    {
      "epoch": 7.283260087894527,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 18230
    },
    {
      "epoch": 7.283659608469836,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 18231
    },
    {
      "epoch": 7.284059129045146,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 18232
    },
    {
      "epoch": 7.284458649620455,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 18233
    },
    {
      "epoch": 7.284858170195765,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 18234
    },
    {
      "epoch": 7.285257690771075,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 18235
    },
    {
      "epoch": 7.285657211346384,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 18236
    },
    {
      "epoch": 7.286056731921694,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 18237
    },
    {
      "epoch": 7.286456252497004,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 18238
    },
    {
      "epoch": 7.286855773072313,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 18239
    },
    {
      "epoch": 7.287255293647623,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 18240
    },
    {
      "epoch": 7.2876548142229325,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 18241
    },
    {
      "epoch": 7.288054334798242,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 18242
    },
    {
      "epoch": 7.288453855373552,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 18243
    },
    {
      "epoch": 7.288853375948861,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 18244
    },
    {
      "epoch": 7.289252896524171,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 18245
    },
    {
      "epoch": 7.2896524170994805,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 18246
    },
    {
      "epoch": 7.29005193767479,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 18247
    },
    {
      "epoch": 7.2904514582501,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 18248
    },
    {
      "epoch": 7.29085097882541,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 18249
    },
    {
      "epoch": 7.291250499400719,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 18250
    },
    {
      "epoch": 7.291650019976029,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 18251
    },
    {
      "epoch": 7.292049540551338,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 18252
    },
    {
      "epoch": 7.292449061126648,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 18253
    },
    {
      "epoch": 7.292848581701958,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 18254
    },
    {
      "epoch": 7.293248102277268,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 18255
    },
    {
      "epoch": 7.293647622852577,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 18256
    },
    {
      "epoch": 7.294047143427886,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 18257
    },
    {
      "epoch": 7.294446664003196,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 18258
    },
    {
      "epoch": 7.294846184578506,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 18259
    },
    {
      "epoch": 7.295245705153816,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 18260
    },
    {
      "epoch": 7.295645225729125,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 18261
    },
    {
      "epoch": 7.2960447463044344,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 18262
    },
    {
      "epoch": 7.296444266879744,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 18263
    },
    {
      "epoch": 7.296843787455054,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 18264
    },
    {
      "epoch": 7.297243308030364,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 18265
    },
    {
      "epoch": 7.297642828605674,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 18266
    },
    {
      "epoch": 7.2980423491809825,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 18267
    },
    {
      "epoch": 7.298441869756292,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 18268
    },
    {
      "epoch": 7.298841390331602,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 18269
    },
    {
      "epoch": 7.299240910906912,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 18270
    },
    {
      "epoch": 7.299640431482222,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 18271
    },
    {
      "epoch": 7.300039952057531,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 18272
    },
    {
      "epoch": 7.30043947263284,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 18273
    },
    {
      "epoch": 7.30083899320815,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 18274
    },
    {
      "epoch": 7.30123851378346,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 18275
    },
    {
      "epoch": 7.30163803435877,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 18276
    },
    {
      "epoch": 7.3020375549340795,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 18277
    },
    {
      "epoch": 7.302437075509388,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 18278
    },
    {
      "epoch": 7.302836596084698,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 18279
    },
    {
      "epoch": 7.303236116660008,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 18280
    },
    {
      "epoch": 7.303635637235318,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 18281
    },
    {
      "epoch": 7.3040351578106275,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 18282
    },
    {
      "epoch": 7.304434678385936,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 18283
    },
    {
      "epoch": 7.304834198961246,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 18284
    },
    {
      "epoch": 7.305233719536556,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 18285
    },
    {
      "epoch": 7.305633240111866,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 18286
    },
    {
      "epoch": 7.3060327606871756,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 18287
    },
    {
      "epoch": 7.306432281262485,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 18288
    },
    {
      "epoch": 7.306831801837794,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 18289
    },
    {
      "epoch": 7.307231322413104,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 18290
    },
    {
      "epoch": 7.307630842988414,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 18291
    },
    {
      "epoch": 7.308030363563724,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 18292
    },
    {
      "epoch": 7.308429884139033,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 18293
    },
    {
      "epoch": 7.308829404714343,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 18294
    },
    {
      "epoch": 7.309228925289652,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 18295
    },
    {
      "epoch": 7.309628445864962,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 18296
    },
    {
      "epoch": 7.310027966440272,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 18297
    },
    {
      "epoch": 7.310427487015581,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 18298
    },
    {
      "epoch": 7.310827007590891,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 18299
    },
    {
      "epoch": 7.3112265281662,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 18300
    },
    {
      "epoch": 7.31162604874151,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 18301
    },
    {
      "epoch": 7.31202556931682,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 18302
    },
    {
      "epoch": 7.3124250898921295,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 18303
    },
    {
      "epoch": 7.312824610467439,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 18304
    },
    {
      "epoch": 7.313224131042749,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 18305
    },
    {
      "epoch": 7.313623651618058,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 18306
    },
    {
      "epoch": 7.314023172193368,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 18307
    },
    {
      "epoch": 7.3144226927686775,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 18308
    },
    {
      "epoch": 7.314822213343987,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 18309
    },
    {
      "epoch": 7.315221733919297,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 18310
    },
    {
      "epoch": 7.315621254494607,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 18311
    },
    {
      "epoch": 7.316020775069916,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 18312
    },
    {
      "epoch": 7.316420295645226,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 18313
    },
    {
      "epoch": 7.316819816220535,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 18314
    },
    {
      "epoch": 7.317219336795845,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 18315
    },
    {
      "epoch": 7.317618857371155,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 18316
    },
    {
      "epoch": 7.318018377946464,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 18317
    },
    {
      "epoch": 7.318417898521774,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 18318
    },
    {
      "epoch": 7.318817419097083,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 18319
    },
    {
      "epoch": 7.319216939672393,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 18320
    },
    {
      "epoch": 7.319616460247703,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 18321
    },
    {
      "epoch": 7.320015980823013,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 18322
    },
    {
      "epoch": 7.320415501398322,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 18323
    },
    {
      "epoch": 7.320815021973631,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 18324
    },
    {
      "epoch": 7.321214542548941,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 18325
    },
    {
      "epoch": 7.321614063124251,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 18326
    },
    {
      "epoch": 7.322013583699561,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 18327
    },
    {
      "epoch": 7.322413104274871,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 18328
    },
    {
      "epoch": 7.3228126248501795,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 18329
    },
    {
      "epoch": 7.323212145425489,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 18330
    },
    {
      "epoch": 7.323611666000799,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 18331
    },
    {
      "epoch": 7.324011186576109,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 18332
    },
    {
      "epoch": 7.324410707151419,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 18333
    },
    {
      "epoch": 7.3248102277267275,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 18334
    },
    {
      "epoch": 7.325209748302037,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 18335
    },
    {
      "epoch": 7.325609268877347,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 18336
    },
    {
      "epoch": 7.326008789452657,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 18337
    },
    {
      "epoch": 7.326408310027967,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 18338
    },
    {
      "epoch": 7.3268078306032765,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 18339
    },
    {
      "epoch": 7.327207351178585,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 18340
    },
    {
      "epoch": 7.327606871753895,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 18341
    },
    {
      "epoch": 7.328006392329205,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 18342
    },
    {
      "epoch": 7.328405912904515,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 18343
    },
    {
      "epoch": 7.3288054334798245,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 18344
    },
    {
      "epoch": 7.329204954055134,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 18345
    },
    {
      "epoch": 7.329604474630443,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 18346
    },
    {
      "epoch": 7.330003995205753,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 18347
    },
    {
      "epoch": 7.330403515781063,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 18348
    },
    {
      "epoch": 7.3308030363563725,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 18349
    },
    {
      "epoch": 7.331202556931682,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 18350
    },
    {
      "epoch": 7.331602077506991,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 18351
    },
    {
      "epoch": 7.332001598082301,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 18352
    },
    {
      "epoch": 7.332401118657611,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 18353
    },
    {
      "epoch": 7.332800639232921,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 18354
    },
    {
      "epoch": 7.33320015980823,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 18355
    },
    {
      "epoch": 7.33359968038354,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 18356
    },
    {
      "epoch": 7.333999200958849,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 18357
    },
    {
      "epoch": 7.334398721534159,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 18358
    },
    {
      "epoch": 7.334798242109469,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 18359
    },
    {
      "epoch": 7.335197762684778,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 18360
    },
    {
      "epoch": 7.335597283260088,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 18361
    },
    {
      "epoch": 7.335996803835398,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 18362
    },
    {
      "epoch": 7.336396324410707,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 18363
    },
    {
      "epoch": 7.336795844986017,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 18364
    },
    {
      "epoch": 7.3371953655613265,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 18365
    },
    {
      "epoch": 7.337594886136636,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 18366
    },
    {
      "epoch": 7.337994406711946,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 18367
    },
    {
      "epoch": 7.338393927287255,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 18368
    },
    {
      "epoch": 7.338793447862565,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 18369
    },
    {
      "epoch": 7.3391929684378745,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 18370
    },
    {
      "epoch": 7.339592489013184,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 18371
    },
    {
      "epoch": 7.339992009588494,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 18372
    },
    {
      "epoch": 7.340391530163803,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 18373
    },
    {
      "epoch": 7.340791050739113,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 18374
    },
    {
      "epoch": 7.3411905713144225,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 18375
    },
    {
      "epoch": 7.341590091889732,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 18376
    },
    {
      "epoch": 7.341989612465042,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 18377
    },
    {
      "epoch": 7.342389133040352,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 18378
    },
    {
      "epoch": 7.342788653615661,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 18379
    },
    {
      "epoch": 7.343188174190971,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 18380
    },
    {
      "epoch": 7.34358769476628,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 18381
    },
    {
      "epoch": 7.34398721534159,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 18382
    },
    {
      "epoch": 7.3443867359169,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 18383
    },
    {
      "epoch": 7.34478625649221,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 18384
    },
    {
      "epoch": 7.345185777067519,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 18385
    },
    {
      "epoch": 7.345585297642828,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 18386
    },
    {
      "epoch": 7.345984818218138,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 18387
    },
    {
      "epoch": 7.346384338793448,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 18388
    },
    {
      "epoch": 7.346783859368758,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 18389
    },
    {
      "epoch": 7.347183379944067,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 18390
    },
    {
      "epoch": 7.3475829005193765,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 18391
    },
    {
      "epoch": 7.347982421094686,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 18392
    },
    {
      "epoch": 7.348381941669996,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 18393
    },
    {
      "epoch": 7.348781462245306,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 18394
    },
    {
      "epoch": 7.349180982820616,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 18395
    },
    {
      "epoch": 7.3495805033959245,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 18396
    },
    {
      "epoch": 7.349980023971234,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 18397
    },
    {
      "epoch": 7.350379544546544,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 18398
    },
    {
      "epoch": 7.350779065121854,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 18399
    },
    {
      "epoch": 7.351178585697164,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 18400
    },
    {
      "epoch": 7.351578106272473,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 18401
    },
    {
      "epoch": 7.351977626847782,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 18402
    },
    {
      "epoch": 7.352377147423092,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 18403
    },
    {
      "epoch": 7.352776667998402,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 18404
    },
    {
      "epoch": 7.353176188573712,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 18405
    },
    {
      "epoch": 7.3535757091490215,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 18406
    },
    {
      "epoch": 7.35397522972433,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 18407
    },
    {
      "epoch": 7.35437475029964,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 18408
    },
    {
      "epoch": 7.35477427087495,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 18409
    },
    {
      "epoch": 7.35517379145026,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 18410
    },
    {
      "epoch": 7.3555733120255695,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 18411
    },
    {
      "epoch": 7.355972832600879,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 18412
    },
    {
      "epoch": 7.356372353176188,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 18413
    },
    {
      "epoch": 7.356771873751498,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 18414
    },
    {
      "epoch": 7.357171394326808,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 18415
    },
    {
      "epoch": 7.357570914902118,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 18416
    },
    {
      "epoch": 7.357970435477427,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 18417
    },
    {
      "epoch": 7.358369956052737,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 18418
    },
    {
      "epoch": 7.358769476628046,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 18419
    },
    {
      "epoch": 7.359168997203356,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 18420
    },
    {
      "epoch": 7.359568517778666,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.052,
      "step": 18421
    },
    {
      "epoch": 7.359968038353975,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 18422
    },
    {
      "epoch": 7.360367558929285,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 18423
    },
    {
      "epoch": 7.360767079504594,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0572,
      "step": 18424
    },
    {
      "epoch": 7.361166600079904,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 18425
    },
    {
      "epoch": 7.361566120655214,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 18426
    },
    {
      "epoch": 7.361965641230523,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 18427
    },
    {
      "epoch": 7.362365161805833,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 18428
    },
    {
      "epoch": 7.362764682381143,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 18429
    },
    {
      "epoch": 7.363164202956452,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 18430
    },
    {
      "epoch": 7.363563723531762,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 18431
    },
    {
      "epoch": 7.3639632441070715,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 18432
    },
    {
      "epoch": 7.364362764682381,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 18433
    },
    {
      "epoch": 7.364762285257691,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 18434
    },
    {
      "epoch": 7.365161805833001,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 18435
    },
    {
      "epoch": 7.36556132640831,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 18436
    },
    {
      "epoch": 7.3659608469836195,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 18437
    },
    {
      "epoch": 7.366360367558929,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 18438
    },
    {
      "epoch": 7.366759888134239,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 18439
    },
    {
      "epoch": 7.367159408709549,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 18440
    },
    {
      "epoch": 7.367558929284858,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 18441
    },
    {
      "epoch": 7.367958449860168,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 18442
    },
    {
      "epoch": 7.368357970435477,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 18443
    },
    {
      "epoch": 7.368757491010787,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 18444
    },
    {
      "epoch": 7.369157011586097,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 18445
    },
    {
      "epoch": 7.369556532161407,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 18446
    },
    {
      "epoch": 7.369956052736716,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 18447
    },
    {
      "epoch": 7.370355573312025,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 18448
    },
    {
      "epoch": 7.370755093887335,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 18449
    },
    {
      "epoch": 7.371154614462645,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 18450
    },
    {
      "epoch": 7.371554135037955,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 18451
    },
    {
      "epoch": 7.3719536556132645,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 18452
    },
    {
      "epoch": 7.372353176188573,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 18453
    },
    {
      "epoch": 7.372752696763883,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 18454
    },
    {
      "epoch": 7.373152217339193,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 18455
    },
    {
      "epoch": 7.373551737914503,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 18456
    },
    {
      "epoch": 7.373951258489813,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 18457
    },
    {
      "epoch": 7.3743507790651215,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 18458
    },
    {
      "epoch": 7.374750299640431,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 18459
    },
    {
      "epoch": 7.375149820215741,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 18460
    },
    {
      "epoch": 7.375549340791051,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 18461
    },
    {
      "epoch": 7.375948861366361,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 18462
    },
    {
      "epoch": 7.3763483819416695,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0609,
      "step": 18463
    },
    {
      "epoch": 7.376747902516979,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 18464
    },
    {
      "epoch": 7.377147423092289,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 18465
    },
    {
      "epoch": 7.377546943667599,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 18466
    },
    {
      "epoch": 7.377946464242909,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 18467
    },
    {
      "epoch": 7.3783459848182185,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 18468
    },
    {
      "epoch": 7.378745505393527,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 18469
    },
    {
      "epoch": 7.379145025968837,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 18470
    },
    {
      "epoch": 7.379544546544147,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 18471
    },
    {
      "epoch": 7.379944067119457,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 18472
    },
    {
      "epoch": 7.3803435876947665,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 18473
    },
    {
      "epoch": 7.380743108270076,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 18474
    },
    {
      "epoch": 7.381142628845385,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 18475
    },
    {
      "epoch": 7.381542149420695,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 18476
    },
    {
      "epoch": 7.381941669996005,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 18477
    },
    {
      "epoch": 7.3823411905713145,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 18478
    },
    {
      "epoch": 7.382740711146624,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 18479
    },
    {
      "epoch": 7.383140231721933,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 18480
    },
    {
      "epoch": 7.383539752297243,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 18481
    },
    {
      "epoch": 7.383939272872553,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 18482
    },
    {
      "epoch": 7.384338793447863,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 18483
    },
    {
      "epoch": 7.384738314023172,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 18484
    },
    {
      "epoch": 7.385137834598482,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 18485
    },
    {
      "epoch": 7.385537355173791,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 18486
    },
    {
      "epoch": 7.385936875749101,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 18487
    },
    {
      "epoch": 7.386336396324411,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 18488
    },
    {
      "epoch": 7.38673591689972,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 18489
    },
    {
      "epoch": 7.38713543747503,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 18490
    },
    {
      "epoch": 7.38753495805034,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 18491
    },
    {
      "epoch": 7.387934478625649,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 18492
    },
    {
      "epoch": 7.388333999200959,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 18493
    },
    {
      "epoch": 7.3887335197762685,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 18494
    },
    {
      "epoch": 7.389133040351578,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 18495
    },
    {
      "epoch": 7.389532560926888,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 18496
    },
    {
      "epoch": 7.389932081502197,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 18497
    },
    {
      "epoch": 7.390331602077507,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 18498
    },
    {
      "epoch": 7.3907311226528165,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 18499
    },
    {
      "epoch": 7.391130643228126,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 18500
    },
    {
      "epoch": 7.391530163803436,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 18501
    },
    {
      "epoch": 7.391929684378746,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 18502
    },
    {
      "epoch": 7.392329204954055,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 18503
    },
    {
      "epoch": 7.3927287255293646,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 18504
    },
    {
      "epoch": 7.393128246104674,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 18505
    },
    {
      "epoch": 7.393527766679984,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 18506
    },
    {
      "epoch": 7.393927287255294,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 18507
    },
    {
      "epoch": 7.394326807830604,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 18508
    },
    {
      "epoch": 7.394726328405913,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 18509
    },
    {
      "epoch": 7.395125848981222,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 18510
    },
    {
      "epoch": 7.395525369556532,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 18511
    },
    {
      "epoch": 7.395924890131842,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 18512
    },
    {
      "epoch": 7.396324410707152,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 18513
    },
    {
      "epoch": 7.396723931282461,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 18514
    },
    {
      "epoch": 7.39712345185777,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 18515
    },
    {
      "epoch": 7.39752297243308,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 18516
    },
    {
      "epoch": 7.39792249300839,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 18517
    },
    {
      "epoch": 7.3983220135837,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 18518
    },
    {
      "epoch": 7.39872153415901,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 18519
    },
    {
      "epoch": 7.3991210547343185,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 18520
    },
    {
      "epoch": 7.399520575309628,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 18521
    },
    {
      "epoch": 7.399920095884938,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 18522
    },
    {
      "epoch": 7.400319616460248,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 18523
    },
    {
      "epoch": 7.400719137035558,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 18524
    },
    {
      "epoch": 7.401118657610867,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 18525
    },
    {
      "epoch": 7.401518178186176,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 18526
    },
    {
      "epoch": 7.401917698761486,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 18527
    },
    {
      "epoch": 7.402317219336796,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 18528
    },
    {
      "epoch": 7.402716739912106,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 18529
    },
    {
      "epoch": 7.4031162604874154,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 18530
    },
    {
      "epoch": 7.403515781062724,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 18531
    },
    {
      "epoch": 7.403915301638034,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 18532
    },
    {
      "epoch": 7.404314822213344,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 18533
    },
    {
      "epoch": 7.404714342788654,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 18534
    },
    {
      "epoch": 7.4051138633639635,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 18535
    },
    {
      "epoch": 7.405513383939273,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 18536
    },
    {
      "epoch": 7.405912904514582,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 18537
    },
    {
      "epoch": 7.406312425089892,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 18538
    },
    {
      "epoch": 7.406711945665202,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 18539
    },
    {
      "epoch": 7.4071114662405115,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 18540
    },
    {
      "epoch": 7.407510986815821,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 18541
    },
    {
      "epoch": 7.407910507391131,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 18542
    },
    {
      "epoch": 7.40831002796644,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 18543
    },
    {
      "epoch": 7.40870954854175,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 18544
    },
    {
      "epoch": 7.40910906911706,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 18545
    },
    {
      "epoch": 7.409508589692369,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 18546
    },
    {
      "epoch": 7.409908110267679,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 18547
    },
    {
      "epoch": 7.410307630842988,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 18548
    },
    {
      "epoch": 7.410707151418298,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 18549
    },
    {
      "epoch": 7.411106671993608,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 18550
    },
    {
      "epoch": 7.411506192568917,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 18551
    },
    {
      "epoch": 7.411905713144227,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 18552
    },
    {
      "epoch": 7.412305233719536,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 18553
    },
    {
      "epoch": 7.412704754294846,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 18554
    },
    {
      "epoch": 7.413104274870156,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 18555
    },
    {
      "epoch": 7.4135037954454654,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 18556
    },
    {
      "epoch": 7.413903316020775,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 18557
    },
    {
      "epoch": 7.414302836596085,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 18558
    },
    {
      "epoch": 7.414702357171394,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 18559
    },
    {
      "epoch": 7.415101877746704,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 18560
    },
    {
      "epoch": 7.4155013983220135,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 18561
    },
    {
      "epoch": 7.415900918897323,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 18562
    },
    {
      "epoch": 7.416300439472633,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 18563
    },
    {
      "epoch": 7.416699960047943,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 18564
    },
    {
      "epoch": 7.417099480623252,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 18565
    },
    {
      "epoch": 7.4174990011985615,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 18566
    },
    {
      "epoch": 7.417898521773871,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 18567
    },
    {
      "epoch": 7.418298042349181,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 18568
    },
    {
      "epoch": 7.418697562924491,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 18569
    },
    {
      "epoch": 7.4190970834998,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 18570
    },
    {
      "epoch": 7.41949660407511,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 18571
    },
    {
      "epoch": 7.419896124650419,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 18572
    },
    {
      "epoch": 7.420295645225729,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 18573
    },
    {
      "epoch": 7.420695165801039,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 18574
    },
    {
      "epoch": 7.421094686376349,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 18575
    },
    {
      "epoch": 7.421494206951658,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 18576
    },
    {
      "epoch": 7.421893727526967,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 18577
    },
    {
      "epoch": 7.422293248102277,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 18578
    },
    {
      "epoch": 7.422692768677587,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 18579
    },
    {
      "epoch": 7.423092289252897,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 18580
    },
    {
      "epoch": 7.4234918098282066,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 18581
    },
    {
      "epoch": 7.4238913304035155,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 18582
    },
    {
      "epoch": 7.424290850978825,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 18583
    },
    {
      "epoch": 7.424690371554135,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 18584
    },
    {
      "epoch": 7.425089892129445,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 18585
    },
    {
      "epoch": 7.425489412704755,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 18586
    },
    {
      "epoch": 7.4258889332800635,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 18587
    },
    {
      "epoch": 7.426288453855373,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 18588
    },
    {
      "epoch": 7.426687974430683,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 18589
    },
    {
      "epoch": 7.427087495005993,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 18590
    },
    {
      "epoch": 7.427487015581303,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 18591
    },
    {
      "epoch": 7.427886536156612,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 18592
    },
    {
      "epoch": 7.428286056731921,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 18593
    },
    {
      "epoch": 7.428685577307231,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 18594
    },
    {
      "epoch": 7.429085097882541,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 18595
    },
    {
      "epoch": 7.429484618457851,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 18596
    },
    {
      "epoch": 7.4298841390331605,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 18597
    },
    {
      "epoch": 7.43028365960847,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 18598
    },
    {
      "epoch": 7.430683180183779,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 18599
    },
    {
      "epoch": 7.431082700759089,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 18600
    },
    {
      "epoch": 7.431482221334399,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 18601
    },
    {
      "epoch": 7.4318817419097085,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 18602
    },
    {
      "epoch": 7.432281262485018,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 18603
    },
    {
      "epoch": 7.432680783060327,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 18604
    },
    {
      "epoch": 7.433080303635637,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 18605
    },
    {
      "epoch": 7.433479824210947,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 18606
    },
    {
      "epoch": 7.433879344786257,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 18607
    },
    {
      "epoch": 7.434278865361566,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 18608
    },
    {
      "epoch": 7.434678385936876,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 18609
    },
    {
      "epoch": 7.435077906512185,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 18610
    },
    {
      "epoch": 7.435477427087495,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 18611
    },
    {
      "epoch": 7.435876947662805,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 18612
    },
    {
      "epoch": 7.436276468238114,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 18613
    },
    {
      "epoch": 7.436675988813424,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 18614
    },
    {
      "epoch": 7.437075509388734,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 18615
    },
    {
      "epoch": 7.437475029964043,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 18616
    },
    {
      "epoch": 7.437874550539353,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 18617
    },
    {
      "epoch": 7.438274071114662,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 18618
    },
    {
      "epoch": 7.438673591689972,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 18619
    },
    {
      "epoch": 7.439073112265282,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 18620
    },
    {
      "epoch": 7.439472632840591,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 18621
    },
    {
      "epoch": 7.439872153415901,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 18622
    },
    {
      "epoch": 7.4402716739912105,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 18623
    },
    {
      "epoch": 7.44067119456652,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 18624
    },
    {
      "epoch": 7.44107071514183,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 18625
    },
    {
      "epoch": 7.44147023571714,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 18626
    },
    {
      "epoch": 7.441869756292449,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 18627
    },
    {
      "epoch": 7.4422692768677585,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 18628
    },
    {
      "epoch": 7.442668797443068,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 18629
    },
    {
      "epoch": 7.443068318018378,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 18630
    },
    {
      "epoch": 7.443467838593688,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 18631
    },
    {
      "epoch": 7.443867359168998,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 18632
    },
    {
      "epoch": 7.444266879744307,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 18633
    },
    {
      "epoch": 7.444666400319616,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 18634
    },
    {
      "epoch": 7.445065920894926,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 18635
    },
    {
      "epoch": 7.445465441470236,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 18636
    },
    {
      "epoch": 7.445864962045546,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 18637
    },
    {
      "epoch": 7.446264482620855,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 18638
    },
    {
      "epoch": 7.446664003196164,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 18639
    },
    {
      "epoch": 7.447063523771474,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 18640
    },
    {
      "epoch": 7.447463044346784,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 18641
    },
    {
      "epoch": 7.447862564922094,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 18642
    },
    {
      "epoch": 7.448262085497403,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 18643
    },
    {
      "epoch": 7.448661606072712,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 18644
    },
    {
      "epoch": 7.449061126648022,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 18645
    },
    {
      "epoch": 7.449460647223332,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 18646
    },
    {
      "epoch": 7.449860167798642,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 18647
    },
    {
      "epoch": 7.450259688373952,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 18648
    },
    {
      "epoch": 7.4506592089492605,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 18649
    },
    {
      "epoch": 7.45105872952457,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 18650
    },
    {
      "epoch": 7.45145825009988,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 18651
    },
    {
      "epoch": 7.45185777067519,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 18652
    },
    {
      "epoch": 7.4522572912505,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 18653
    },
    {
      "epoch": 7.452656811825809,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 18654
    },
    {
      "epoch": 7.453056332401118,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 18655
    },
    {
      "epoch": 7.453455852976428,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 18656
    },
    {
      "epoch": 7.453855373551738,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0553,
      "step": 18657
    },
    {
      "epoch": 7.454254894127048,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 18658
    },
    {
      "epoch": 7.4546544147023575,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 18659
    },
    {
      "epoch": 7.455053935277666,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 18660
    },
    {
      "epoch": 7.455453455852976,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 18661
    },
    {
      "epoch": 7.455852976428286,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 18662
    },
    {
      "epoch": 7.456252497003596,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 18663
    },
    {
      "epoch": 7.4566520175789055,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 18664
    },
    {
      "epoch": 7.457051538154215,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 18665
    },
    {
      "epoch": 7.457451058729524,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 18666
    },
    {
      "epoch": 7.457850579304834,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 18667
    },
    {
      "epoch": 7.458250099880144,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 18668
    },
    {
      "epoch": 7.4586496204554535,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 18669
    },
    {
      "epoch": 7.459049141030763,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 18670
    },
    {
      "epoch": 7.459448661606073,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 18671
    },
    {
      "epoch": 7.459848182181382,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 18672
    },
    {
      "epoch": 7.460247702756692,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 18673
    },
    {
      "epoch": 7.460647223332002,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 18674
    },
    {
      "epoch": 7.461046743907311,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 18675
    },
    {
      "epoch": 7.461446264482621,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 18676
    },
    {
      "epoch": 7.46184578505793,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 18677
    },
    {
      "epoch": 7.46224530563324,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 18678
    },
    {
      "epoch": 7.46264482620855,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 18679
    },
    {
      "epoch": 7.463044346783859,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 18680
    },
    {
      "epoch": 7.463443867359169,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 18681
    },
    {
      "epoch": 7.463843387934479,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 18682
    },
    {
      "epoch": 7.464242908509788,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 18683
    },
    {
      "epoch": 7.464642429085098,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 18684
    },
    {
      "epoch": 7.4650419496604075,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 18685
    },
    {
      "epoch": 7.465441470235717,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 18686
    },
    {
      "epoch": 7.465840990811027,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 18687
    },
    {
      "epoch": 7.466240511386337,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 18688
    },
    {
      "epoch": 7.466640031961646,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 18689
    },
    {
      "epoch": 7.4670395525369555,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 18690
    },
    {
      "epoch": 7.467439073112265,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 18691
    },
    {
      "epoch": 7.467838593687575,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 18692
    },
    {
      "epoch": 7.468238114262885,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 18693
    },
    {
      "epoch": 7.468637634838194,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 18694
    },
    {
      "epoch": 7.4690371554135035,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 18695
    },
    {
      "epoch": 7.469436675988813,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 18696
    },
    {
      "epoch": 7.469836196564123,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 18697
    },
    {
      "epoch": 7.470235717139433,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 18698
    },
    {
      "epoch": 7.470635237714743,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 18699
    },
    {
      "epoch": 7.471034758290052,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 18700
    },
    {
      "epoch": 7.471434278865361,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 18701
    },
    {
      "epoch": 7.471833799440671,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 18702
    },
    {
      "epoch": 7.472233320015981,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 18703
    },
    {
      "epoch": 7.472632840591291,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 18704
    },
    {
      "epoch": 7.4730323611666005,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 18705
    },
    {
      "epoch": 7.473431881741909,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 18706
    },
    {
      "epoch": 7.473831402317219,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 18707
    },
    {
      "epoch": 7.474230922892529,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 18708
    },
    {
      "epoch": 7.474630443467839,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 18709
    },
    {
      "epoch": 7.475029964043149,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 18710
    },
    {
      "epoch": 7.4754294846184575,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 18711
    },
    {
      "epoch": 7.475829005193767,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 18712
    },
    {
      "epoch": 7.476228525769077,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 18713
    },
    {
      "epoch": 7.476628046344387,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 18714
    },
    {
      "epoch": 7.477027566919697,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 18715
    },
    {
      "epoch": 7.477427087495006,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 18716
    },
    {
      "epoch": 7.477826608070315,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 18717
    },
    {
      "epoch": 7.478226128645625,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 18718
    },
    {
      "epoch": 7.478625649220935,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 18719
    },
    {
      "epoch": 7.479025169796245,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 18720
    },
    {
      "epoch": 7.479424690371554,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 18721
    },
    {
      "epoch": 7.479824210946864,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 18722
    },
    {
      "epoch": 7.480223731522173,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 18723
    },
    {
      "epoch": 7.480623252097483,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 18724
    },
    {
      "epoch": 7.481022772672793,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 18725
    },
    {
      "epoch": 7.4814222932481025,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 18726
    },
    {
      "epoch": 7.481821813823412,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 18727
    },
    {
      "epoch": 7.482221334398721,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 18728
    },
    {
      "epoch": 7.482620854974031,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 18729
    },
    {
      "epoch": 7.483020375549341,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 18730
    },
    {
      "epoch": 7.4834198961246505,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 18731
    },
    {
      "epoch": 7.48381941669996,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 18732
    },
    {
      "epoch": 7.484218937275269,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 18733
    },
    {
      "epoch": 7.484618457850579,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 18734
    },
    {
      "epoch": 7.485017978425889,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 18735
    },
    {
      "epoch": 7.485417499001199,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 18736
    },
    {
      "epoch": 7.485817019576508,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 18737
    },
    {
      "epoch": 7.486216540151818,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.056,
      "step": 18738
    },
    {
      "epoch": 7.486616060727127,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 18739
    },
    {
      "epoch": 7.487015581302437,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 18740
    },
    {
      "epoch": 7.487415101877747,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 18741
    },
    {
      "epoch": 7.487814622453056,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 18742
    },
    {
      "epoch": 7.488214143028366,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 18743
    },
    {
      "epoch": 7.488613663603676,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 18744
    },
    {
      "epoch": 7.489013184178985,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 18745
    },
    {
      "epoch": 7.489412704754295,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 18746
    },
    {
      "epoch": 7.4898122253296044,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 18747
    },
    {
      "epoch": 7.490211745904914,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 18748
    },
    {
      "epoch": 7.490611266480224,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 18749
    },
    {
      "epoch": 7.491010787055533,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 18750
    },
    {
      "epoch": 7.491410307630843,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 18751
    },
    {
      "epoch": 7.4918098282061525,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 18752
    },
    {
      "epoch": 7.492209348781462,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 18753
    },
    {
      "epoch": 7.492608869356772,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 18754
    },
    {
      "epoch": 7.493008389932082,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 18755
    },
    {
      "epoch": 7.493407910507391,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 18756
    },
    {
      "epoch": 7.4938074310827005,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 18757
    },
    {
      "epoch": 7.49420695165801,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 18758
    },
    {
      "epoch": 7.49460647223332,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 18759
    },
    {
      "epoch": 7.49500599280863,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 18760
    },
    {
      "epoch": 7.49540551338394,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 18761
    },
    {
      "epoch": 7.495805033959249,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 18762
    },
    {
      "epoch": 7.496204554534558,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 18763
    },
    {
      "epoch": 7.496604075109868,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 18764
    },
    {
      "epoch": 7.497003595685178,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 18765
    },
    {
      "epoch": 7.497403116260488,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 18766
    },
    {
      "epoch": 7.497802636835797,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 18767
    },
    {
      "epoch": 7.498202157411106,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 18768
    },
    {
      "epoch": 7.498601677986416,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 18769
    },
    {
      "epoch": 7.499001198561726,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 18770
    },
    {
      "epoch": 7.499400719137036,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 18771
    },
    {
      "epoch": 7.4998002397123456,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 18772
    },
    {
      "epoch": 7.5001997602876544,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 18773
    },
    {
      "epoch": 7.500599280862964,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 18774
    },
    {
      "epoch": 7.500998801438274,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 18775
    },
    {
      "epoch": 7.501398322013584,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 18776
    },
    {
      "epoch": 7.501797842588894,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 18777
    },
    {
      "epoch": 7.502197363164203,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 18778
    },
    {
      "epoch": 7.502596883739512,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 18779
    },
    {
      "epoch": 7.502996404314822,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 18780
    },
    {
      "epoch": 7.503395924890132,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 18781
    },
    {
      "epoch": 7.503795445465442,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 18782
    },
    {
      "epoch": 7.504194966040751,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 18783
    },
    {
      "epoch": 7.50459448661606,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 18784
    },
    {
      "epoch": 7.50499400719137,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 18785
    },
    {
      "epoch": 7.50539352776668,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 18786
    },
    {
      "epoch": 7.50579304834199,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 18787
    },
    {
      "epoch": 7.5061925689172995,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 18788
    },
    {
      "epoch": 7.506592089492609,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 18789
    },
    {
      "epoch": 7.506991610067918,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 18790
    },
    {
      "epoch": 7.507391130643228,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 18791
    },
    {
      "epoch": 7.507790651218538,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 18792
    },
    {
      "epoch": 7.5081901717938475,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 18793
    },
    {
      "epoch": 7.508589692369157,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 18794
    },
    {
      "epoch": 7.508989212944467,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 18795
    },
    {
      "epoch": 7.509388733519776,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 18796
    },
    {
      "epoch": 7.509788254095086,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 18797
    },
    {
      "epoch": 7.5101877746703956,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 18798
    },
    {
      "epoch": 7.510587295245705,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 18799
    },
    {
      "epoch": 7.510986815821015,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 18800
    },
    {
      "epoch": 7.511386336396324,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 18801
    },
    {
      "epoch": 7.511785856971634,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 18802
    },
    {
      "epoch": 7.512185377546944,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 18803
    },
    {
      "epoch": 7.512584898122253,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 18804
    },
    {
      "epoch": 7.512984418697563,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 18805
    },
    {
      "epoch": 7.513383939272872,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 18806
    },
    {
      "epoch": 7.513783459848182,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 18807
    },
    {
      "epoch": 7.514182980423492,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 18808
    },
    {
      "epoch": 7.514582500998801,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 18809
    },
    {
      "epoch": 7.514982021574111,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 18810
    },
    {
      "epoch": 7.515381542149421,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 18811
    },
    {
      "epoch": 7.515781062724731,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 18812
    },
    {
      "epoch": 7.51618058330004,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 18813
    },
    {
      "epoch": 7.5165801038753495,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 18814
    },
    {
      "epoch": 7.516979624450659,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 18815
    },
    {
      "epoch": 7.517379145025969,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 18816
    },
    {
      "epoch": 7.517778665601279,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 18817
    },
    {
      "epoch": 7.518178186176588,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 18818
    },
    {
      "epoch": 7.5185777067518975,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 18819
    },
    {
      "epoch": 7.518977227327207,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 18820
    },
    {
      "epoch": 7.519376747902517,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 18821
    },
    {
      "epoch": 7.519776268477827,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 18822
    },
    {
      "epoch": 7.520175789053136,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 18823
    },
    {
      "epoch": 7.520575309628446,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 18824
    },
    {
      "epoch": 7.520974830203755,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 18825
    },
    {
      "epoch": 7.521374350779065,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 18826
    },
    {
      "epoch": 7.521773871354375,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 18827
    },
    {
      "epoch": 7.522173391929685,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 18828
    },
    {
      "epoch": 7.5225729125049945,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 18829
    },
    {
      "epoch": 7.522972433080303,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 18830
    },
    {
      "epoch": 7.523371953655613,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 18831
    },
    {
      "epoch": 7.523771474230923,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 18832
    },
    {
      "epoch": 7.524170994806233,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 18833
    },
    {
      "epoch": 7.5245705153815425,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 18834
    },
    {
      "epoch": 7.524970035956851,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 18835
    },
    {
      "epoch": 7.525369556532161,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 18836
    },
    {
      "epoch": 7.525769077107471,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 18837
    },
    {
      "epoch": 7.526168597682781,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 18838
    },
    {
      "epoch": 7.526568118258091,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 18839
    },
    {
      "epoch": 7.5269676388333995,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 18840
    },
    {
      "epoch": 7.527367159408709,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 18841
    },
    {
      "epoch": 7.527766679984019,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 18842
    },
    {
      "epoch": 7.528166200559329,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 18843
    },
    {
      "epoch": 7.528565721134639,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 18844
    },
    {
      "epoch": 7.528965241709948,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 18845
    },
    {
      "epoch": 7.529364762285257,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 18846
    },
    {
      "epoch": 7.529764282860567,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 18847
    },
    {
      "epoch": 7.530163803435877,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 18848
    },
    {
      "epoch": 7.530563324011187,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 18849
    },
    {
      "epoch": 7.5309628445864965,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 18850
    },
    {
      "epoch": 7.531362365161806,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 18851
    },
    {
      "epoch": 7.531761885737115,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 18852
    },
    {
      "epoch": 7.532161406312425,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 18853
    },
    {
      "epoch": 7.532560926887735,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 18854
    },
    {
      "epoch": 7.5329604474630445,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 18855
    },
    {
      "epoch": 7.533359968038354,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 18856
    },
    {
      "epoch": 7.533759488613663,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 18857
    },
    {
      "epoch": 7.534159009188973,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 18858
    },
    {
      "epoch": 7.534558529764283,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 18859
    },
    {
      "epoch": 7.5349580503395925,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 18860
    },
    {
      "epoch": 7.535357570914902,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 18861
    },
    {
      "epoch": 7.535757091490212,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 18862
    },
    {
      "epoch": 7.536156612065521,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 18863
    },
    {
      "epoch": 7.536556132640831,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 18864
    },
    {
      "epoch": 7.536955653216141,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 18865
    },
    {
      "epoch": 7.53735517379145,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 18866
    },
    {
      "epoch": 7.53775469436676,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 18867
    },
    {
      "epoch": 7.53815421494207,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 18868
    },
    {
      "epoch": 7.538553735517379,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 18869
    },
    {
      "epoch": 7.538953256092689,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 18870
    },
    {
      "epoch": 7.539352776667998,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 18871
    },
    {
      "epoch": 7.539752297243308,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 18872
    },
    {
      "epoch": 7.540151817818618,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 18873
    },
    {
      "epoch": 7.540551338393927,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 18874
    },
    {
      "epoch": 7.540950858969237,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 18875
    },
    {
      "epoch": 7.5413503795445465,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 18876
    },
    {
      "epoch": 7.541749900119856,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 18877
    },
    {
      "epoch": 7.542149420695166,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 18878
    },
    {
      "epoch": 7.542548941270476,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 18879
    },
    {
      "epoch": 7.542948461845785,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 18880
    },
    {
      "epoch": 7.5433479824210945,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 18881
    },
    {
      "epoch": 7.543747502996404,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 18882
    },
    {
      "epoch": 7.544147023571714,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 18883
    },
    {
      "epoch": 7.544546544147024,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 18884
    },
    {
      "epoch": 7.544946064722334,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 18885
    },
    {
      "epoch": 7.5453455852976425,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 18886
    },
    {
      "epoch": 7.545745105872952,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 18887
    },
    {
      "epoch": 7.546144626448262,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 18888
    },
    {
      "epoch": 7.546544147023572,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 18889
    },
    {
      "epoch": 7.546943667598882,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 18890
    },
    {
      "epoch": 7.547343188174191,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 18891
    },
    {
      "epoch": 7.5477427087495,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 18892
    },
    {
      "epoch": 7.54814222932481,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 18893
    },
    {
      "epoch": 7.54854174990012,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 18894
    },
    {
      "epoch": 7.54894127047543,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 18895
    },
    {
      "epoch": 7.549340791050739,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 18896
    },
    {
      "epoch": 7.549740311626048,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 18897
    },
    {
      "epoch": 7.550139832201358,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 18898
    },
    {
      "epoch": 7.550539352776668,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 18899
    },
    {
      "epoch": 7.550938873351978,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 18900
    },
    {
      "epoch": 7.551338393927288,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 18901
    },
    {
      "epoch": 7.551737914502597,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 18902
    },
    {
      "epoch": 7.552137435077906,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 18903
    },
    {
      "epoch": 7.552536955653216,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 18904
    },
    {
      "epoch": 7.552936476228526,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 18905
    },
    {
      "epoch": 7.553335996803836,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 18906
    },
    {
      "epoch": 7.553735517379145,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 18907
    },
    {
      "epoch": 7.554135037954454,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 18908
    },
    {
      "epoch": 7.554534558529764,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 18909
    },
    {
      "epoch": 7.554934079105074,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 18910
    },
    {
      "epoch": 7.555333599680384,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 18911
    },
    {
      "epoch": 7.555733120255693,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 18912
    },
    {
      "epoch": 7.556132640831002,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 18913
    },
    {
      "epoch": 7.556532161406312,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 18914
    },
    {
      "epoch": 7.556931681981622,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 18915
    },
    {
      "epoch": 7.557331202556932,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 18916
    },
    {
      "epoch": 7.5577307231322415,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 18917
    },
    {
      "epoch": 7.558130243707551,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 18918
    },
    {
      "epoch": 7.558529764282861,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 18919
    },
    {
      "epoch": 7.55892928485817,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 18920
    },
    {
      "epoch": 7.55932880543348,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 18921
    },
    {
      "epoch": 7.5597283260087895,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 18922
    },
    {
      "epoch": 7.560127846584099,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 18923
    },
    {
      "epoch": 7.560527367159409,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 18924
    },
    {
      "epoch": 7.560926887734718,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 18925
    },
    {
      "epoch": 7.561326408310028,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 18926
    },
    {
      "epoch": 7.561725928885338,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 18927
    },
    {
      "epoch": 7.562125449460647,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 18928
    },
    {
      "epoch": 7.562524970035957,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 18929
    },
    {
      "epoch": 7.562924490611266,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 18930
    },
    {
      "epoch": 7.563324011186576,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 18931
    },
    {
      "epoch": 7.563723531761886,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 18932
    },
    {
      "epoch": 7.564123052337195,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 18933
    },
    {
      "epoch": 7.564522572912505,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 18934
    },
    {
      "epoch": 7.564922093487815,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 18935
    },
    {
      "epoch": 7.565321614063124,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 18936
    },
    {
      "epoch": 7.565721134638434,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 18937
    },
    {
      "epoch": 7.566120655213743,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 18938
    },
    {
      "epoch": 7.566520175789053,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 18939
    },
    {
      "epoch": 7.566919696364363,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 18940
    },
    {
      "epoch": 7.567319216939673,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 18941
    },
    {
      "epoch": 7.567718737514982,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 18942
    },
    {
      "epoch": 7.5681182580902915,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 18943
    },
    {
      "epoch": 7.568517778665601,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 18944
    },
    {
      "epoch": 7.568917299240911,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 18945
    },
    {
      "epoch": 7.569316819816221,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 18946
    },
    {
      "epoch": 7.56971634039153,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 18947
    },
    {
      "epoch": 7.5701158609668395,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 18948
    },
    {
      "epoch": 7.570515381542149,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 18949
    },
    {
      "epoch": 7.570914902117459,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 18950
    },
    {
      "epoch": 7.571314422692769,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 18951
    },
    {
      "epoch": 7.571713943268079,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 18952
    },
    {
      "epoch": 7.572113463843388,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 18953
    },
    {
      "epoch": 7.572512984418697,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 18954
    },
    {
      "epoch": 7.572912504994007,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 18955
    },
    {
      "epoch": 7.573312025569317,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 18956
    },
    {
      "epoch": 7.573711546144627,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 18957
    },
    {
      "epoch": 7.5741110667199365,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 18958
    },
    {
      "epoch": 7.574510587295245,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 18959
    },
    {
      "epoch": 7.574910107870555,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 18960
    },
    {
      "epoch": 7.575309628445865,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 18961
    },
    {
      "epoch": 7.575709149021175,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 18962
    },
    {
      "epoch": 7.5761086695964845,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 18963
    },
    {
      "epoch": 7.5765081901717934,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 18964
    },
    {
      "epoch": 7.576907710747103,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 18965
    },
    {
      "epoch": 7.577307231322413,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 18966
    },
    {
      "epoch": 7.577706751897723,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 18967
    },
    {
      "epoch": 7.578106272473033,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 18968
    },
    {
      "epoch": 7.578505793048342,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 18969
    },
    {
      "epoch": 7.578905313623651,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 18970
    },
    {
      "epoch": 7.579304834198961,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 18971
    },
    {
      "epoch": 7.579704354774271,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 18972
    },
    {
      "epoch": 7.580103875349581,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 18973
    },
    {
      "epoch": 7.58050339592489,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 18974
    },
    {
      "epoch": 7.5809029165002,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 18975
    },
    {
      "epoch": 7.581302437075509,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 18976
    },
    {
      "epoch": 7.581701957650819,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 18977
    },
    {
      "epoch": 7.582101478226129,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 18978
    },
    {
      "epoch": 7.5825009988014385,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 18979
    },
    {
      "epoch": 7.582900519376748,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 18980
    },
    {
      "epoch": 7.583300039952057,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 18981
    },
    {
      "epoch": 7.583699560527367,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 18982
    },
    {
      "epoch": 7.584099081102677,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 18983
    },
    {
      "epoch": 7.5844986016779865,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 18984
    },
    {
      "epoch": 7.584898122253296,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 18985
    },
    {
      "epoch": 7.585297642828605,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 18986
    },
    {
      "epoch": 7.585697163403915,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 18987
    },
    {
      "epoch": 7.586096683979225,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 18988
    },
    {
      "epoch": 7.5864962045545346,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 18989
    },
    {
      "epoch": 7.586895725129844,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 18990
    },
    {
      "epoch": 7.587295245705154,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 18991
    },
    {
      "epoch": 7.587694766280464,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 18992
    },
    {
      "epoch": 7.588094286855773,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 18993
    },
    {
      "epoch": 7.588493807431083,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 18994
    },
    {
      "epoch": 7.588893328006392,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 18995
    },
    {
      "epoch": 7.589292848581702,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 18996
    },
    {
      "epoch": 7.589692369157012,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 18997
    },
    {
      "epoch": 7.590091889732321,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 18998
    },
    {
      "epoch": 7.590491410307631,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 18999
    },
    {
      "epoch": 7.59089093088294,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 19000
    },
    {
      "epoch": 7.59129045145825,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 19001
    },
    {
      "epoch": 7.59168997203356,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 19002
    },
    {
      "epoch": 7.592089492608869,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 19003
    },
    {
      "epoch": 7.592489013184179,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 19004
    },
    {
      "epoch": 7.5928885337594885,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 19005
    },
    {
      "epoch": 7.593288054334798,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 19006
    },
    {
      "epoch": 7.593687574910108,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 19007
    },
    {
      "epoch": 7.594087095485418,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 19008
    },
    {
      "epoch": 7.594486616060728,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 19009
    },
    {
      "epoch": 7.5948861366360365,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 19010
    },
    {
      "epoch": 7.595285657211346,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 19011
    },
    {
      "epoch": 7.595685177786656,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 19012
    },
    {
      "epoch": 7.596084698361966,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 19013
    },
    {
      "epoch": 7.596484218937276,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 19014
    },
    {
      "epoch": 7.5968837395125846,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 19015
    },
    {
      "epoch": 7.597283260087894,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 19016
    },
    {
      "epoch": 7.597682780663204,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 19017
    },
    {
      "epoch": 7.598082301238514,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 19018
    },
    {
      "epoch": 7.598481821813824,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 19019
    },
    {
      "epoch": 7.598881342389133,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 19020
    },
    {
      "epoch": 7.599280862964442,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 19021
    },
    {
      "epoch": 7.599680383539752,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 19022
    },
    {
      "epoch": 7.600079904115062,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 19023
    },
    {
      "epoch": 7.600479424690372,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 19024
    },
    {
      "epoch": 7.6008789452656815,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 19025
    },
    {
      "epoch": 7.60127846584099,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 19026
    },
    {
      "epoch": 7.6016779864163,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 19027
    },
    {
      "epoch": 7.60207750699161,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 19028
    },
    {
      "epoch": 7.60247702756692,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 19029
    },
    {
      "epoch": 7.60287654814223,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 19030
    },
    {
      "epoch": 7.603276068717539,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 19031
    },
    {
      "epoch": 7.603675589292848,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 19032
    },
    {
      "epoch": 7.604075109868158,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 19033
    },
    {
      "epoch": 7.604474630443468,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 19034
    },
    {
      "epoch": 7.604874151018778,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 19035
    },
    {
      "epoch": 7.605273671594087,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 19036
    },
    {
      "epoch": 7.605673192169396,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 19037
    },
    {
      "epoch": 7.606072712744706,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 19038
    },
    {
      "epoch": 7.606472233320016,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 19039
    },
    {
      "epoch": 7.606871753895326,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 19040
    },
    {
      "epoch": 7.6072712744706354,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 19041
    },
    {
      "epoch": 7.607670795045945,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 19042
    },
    {
      "epoch": 7.608070315621254,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 19043
    },
    {
      "epoch": 7.608469836196564,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 19044
    },
    {
      "epoch": 7.608869356771874,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 19045
    },
    {
      "epoch": 7.6092688773471835,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 19046
    },
    {
      "epoch": 7.609668397922493,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 19047
    },
    {
      "epoch": 7.610067918497803,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 19048
    },
    {
      "epoch": 7.610467439073112,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 19049
    },
    {
      "epoch": 7.610866959648422,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 19050
    },
    {
      "epoch": 7.6112664802237315,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 19051
    },
    {
      "epoch": 7.611666000799041,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 19052
    },
    {
      "epoch": 7.612065521374351,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 19053
    },
    {
      "epoch": 7.61246504194966,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 19054
    },
    {
      "epoch": 7.61286456252497,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 19055
    },
    {
      "epoch": 7.61326408310028,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 19056
    },
    {
      "epoch": 7.613663603675589,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 19057
    },
    {
      "epoch": 7.614063124250899,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 19058
    },
    {
      "epoch": 7.614462644826209,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 19059
    },
    {
      "epoch": 7.614862165401518,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 19060
    },
    {
      "epoch": 7.615261685976828,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 19061
    },
    {
      "epoch": 7.615661206552137,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 19062
    },
    {
      "epoch": 7.616060727127447,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 19063
    },
    {
      "epoch": 7.616460247702757,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 19064
    },
    {
      "epoch": 7.616859768278067,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 19065
    },
    {
      "epoch": 7.617259288853376,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 19066
    },
    {
      "epoch": 7.6176588094286855,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 19067
    },
    {
      "epoch": 7.618058330003995,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 19068
    },
    {
      "epoch": 7.618457850579305,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 19069
    },
    {
      "epoch": 7.618857371154615,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 19070
    },
    {
      "epoch": 7.619256891729924,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 19071
    },
    {
      "epoch": 7.6196564123052335,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 19072
    },
    {
      "epoch": 7.620055932880543,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 19073
    },
    {
      "epoch": 7.620455453455853,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 19074
    },
    {
      "epoch": 7.620854974031163,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 19075
    },
    {
      "epoch": 7.621254494606472,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 19076
    },
    {
      "epoch": 7.6216540151817815,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 19077
    },
    {
      "epoch": 7.622053535757091,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 19078
    },
    {
      "epoch": 7.622453056332401,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 19079
    },
    {
      "epoch": 7.622852576907711,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 19080
    },
    {
      "epoch": 7.623252097483021,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 19081
    },
    {
      "epoch": 7.6236516180583305,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 19082
    },
    {
      "epoch": 7.624051138633639,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 19083
    },
    {
      "epoch": 7.624450659208949,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 19084
    },
    {
      "epoch": 7.624850179784259,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 19085
    },
    {
      "epoch": 7.625249700359569,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 19086
    },
    {
      "epoch": 7.6256492209348785,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 19087
    },
    {
      "epoch": 7.626048741510187,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 19088
    },
    {
      "epoch": 7.626448262085497,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 19089
    },
    {
      "epoch": 7.626847782660807,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 19090
    },
    {
      "epoch": 7.627247303236117,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 19091
    },
    {
      "epoch": 7.627646823811427,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 19092
    },
    {
      "epoch": 7.6280463443867355,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 19093
    },
    {
      "epoch": 7.628445864962045,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 19094
    },
    {
      "epoch": 7.628845385537355,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 19095
    },
    {
      "epoch": 7.629244906112665,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 19096
    },
    {
      "epoch": 7.629644426687975,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 19097
    },
    {
      "epoch": 7.630043947263284,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 19098
    },
    {
      "epoch": 7.630443467838594,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 19099
    },
    {
      "epoch": 7.630842988413903,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 19100
    },
    {
      "epoch": 7.631242508989213,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 19101
    },
    {
      "epoch": 7.631642029564523,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 19102
    },
    {
      "epoch": 7.632041550139832,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 19103
    },
    {
      "epoch": 7.632441070715142,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 19104
    },
    {
      "epoch": 7.632840591290451,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 19105
    },
    {
      "epoch": 7.633240111865761,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 19106
    },
    {
      "epoch": 7.633639632441071,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 19107
    },
    {
      "epoch": 7.6340391530163805,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 19108
    },
    {
      "epoch": 7.63443867359169,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 19109
    },
    {
      "epoch": 7.634838194166999,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 19110
    },
    {
      "epoch": 7.635237714742309,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 19111
    },
    {
      "epoch": 7.635637235317619,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 19112
    },
    {
      "epoch": 7.6360367558929285,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 19113
    },
    {
      "epoch": 7.636436276468238,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 19114
    },
    {
      "epoch": 7.636835797043548,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 19115
    },
    {
      "epoch": 7.637235317618857,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 19116
    },
    {
      "epoch": 7.637634838194167,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 19117
    },
    {
      "epoch": 7.638034358769477,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 19118
    },
    {
      "epoch": 7.638433879344786,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 19119
    },
    {
      "epoch": 7.638833399920096,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 19120
    },
    {
      "epoch": 7.639232920495406,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 19121
    },
    {
      "epoch": 7.639632441070715,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 19122
    },
    {
      "epoch": 7.640031961646025,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 19123
    },
    {
      "epoch": 7.640431482221334,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 19124
    },
    {
      "epoch": 7.640831002796644,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 19125
    },
    {
      "epoch": 7.641230523371954,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 19126
    },
    {
      "epoch": 7.641630043947263,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 19127
    },
    {
      "epoch": 7.642029564522573,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 19128
    },
    {
      "epoch": 7.642429085097882,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 19129
    },
    {
      "epoch": 7.642828605673192,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 19130
    },
    {
      "epoch": 7.643228126248502,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 19131
    },
    {
      "epoch": 7.643627646823812,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 19132
    },
    {
      "epoch": 7.644027167399121,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 19133
    },
    {
      "epoch": 7.6444266879744305,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 19134
    },
    {
      "epoch": 7.64482620854974,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 19135
    },
    {
      "epoch": 7.64522572912505,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 19136
    },
    {
      "epoch": 7.64562524970036,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 19137
    },
    {
      "epoch": 7.64602477027567,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 19138
    },
    {
      "epoch": 7.6464242908509785,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 19139
    },
    {
      "epoch": 7.646823811426288,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 19140
    },
    {
      "epoch": 7.647223332001598,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 19141
    },
    {
      "epoch": 7.647622852576908,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 19142
    },
    {
      "epoch": 7.648022373152218,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 19143
    },
    {
      "epoch": 7.648421893727527,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 19144
    },
    {
      "epoch": 7.648821414302836,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 19145
    },
    {
      "epoch": 7.649220934878146,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 19146
    },
    {
      "epoch": 7.649620455453456,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 19147
    },
    {
      "epoch": 7.650019976028766,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 19148
    },
    {
      "epoch": 7.6504194966040755,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 19149
    },
    {
      "epoch": 7.650819017179384,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 19150
    },
    {
      "epoch": 7.651218537754694,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 19151
    },
    {
      "epoch": 7.651618058330004,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 19152
    },
    {
      "epoch": 7.652017578905314,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 19153
    },
    {
      "epoch": 7.6524170994806235,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 19154
    },
    {
      "epoch": 7.652816620055933,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 19155
    },
    {
      "epoch": 7.653216140631242,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 19156
    },
    {
      "epoch": 7.653615661206552,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 19157
    },
    {
      "epoch": 7.654015181781862,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 19158
    },
    {
      "epoch": 7.654414702357172,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 19159
    },
    {
      "epoch": 7.654814222932481,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 19160
    },
    {
      "epoch": 7.65521374350779,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 19161
    },
    {
      "epoch": 7.6556132640831,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 19162
    },
    {
      "epoch": 7.65601278465841,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 19163
    },
    {
      "epoch": 7.65641230523372,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 19164
    },
    {
      "epoch": 7.656811825809029,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 19165
    },
    {
      "epoch": 7.657211346384338,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 19166
    },
    {
      "epoch": 7.657610866959648,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 19167
    },
    {
      "epoch": 7.658010387534958,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 19168
    },
    {
      "epoch": 7.658409908110268,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 19169
    },
    {
      "epoch": 7.6588094286855775,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 19170
    },
    {
      "epoch": 7.659208949260887,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 19171
    },
    {
      "epoch": 7.659608469836197,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 19172
    },
    {
      "epoch": 7.660007990411506,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 19173
    },
    {
      "epoch": 7.660407510986816,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 19174
    },
    {
      "epoch": 7.6608070315621255,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0545,
      "step": 19175
    },
    {
      "epoch": 7.661206552137435,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 19176
    },
    {
      "epoch": 7.661606072712745,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 19177
    },
    {
      "epoch": 7.662005593288054,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 19178
    },
    {
      "epoch": 7.662405113863364,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 19179
    },
    {
      "epoch": 7.6628046344386735,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 19180
    },
    {
      "epoch": 7.663204155013983,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 19181
    },
    {
      "epoch": 7.663603675589293,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 19182
    },
    {
      "epoch": 7.664003196164602,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 19183
    },
    {
      "epoch": 7.664402716739912,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 19184
    },
    {
      "epoch": 7.664802237315222,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 19185
    },
    {
      "epoch": 7.665201757890531,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 19186
    },
    {
      "epoch": 7.665601278465841,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 19187
    },
    {
      "epoch": 7.666000799041151,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 19188
    },
    {
      "epoch": 7.666400319616461,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 19189
    },
    {
      "epoch": 7.66679984019177,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 19190
    },
    {
      "epoch": 7.667199360767079,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 19191
    },
    {
      "epoch": 7.667598881342389,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 19192
    },
    {
      "epoch": 7.667998401917699,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 19193
    },
    {
      "epoch": 7.668397922493009,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 19194
    },
    {
      "epoch": 7.668797443068318,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 19195
    },
    {
      "epoch": 7.6691969636436275,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 19196
    },
    {
      "epoch": 7.669596484218937,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 19197
    },
    {
      "epoch": 7.669996004794247,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 19198
    },
    {
      "epoch": 7.670395525369557,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 19199
    },
    {
      "epoch": 7.670795045944866,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 19200
    },
    {
      "epoch": 7.6711945665201755,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 19201
    },
    {
      "epoch": 7.671594087095485,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 19202
    },
    {
      "epoch": 7.671993607670795,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 19203
    },
    {
      "epoch": 7.672393128246105,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 19204
    },
    {
      "epoch": 7.672792648821415,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 19205
    },
    {
      "epoch": 7.6731921693967235,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 19206
    },
    {
      "epoch": 7.673591689972033,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 19207
    },
    {
      "epoch": 7.673991210547343,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 19208
    },
    {
      "epoch": 7.674390731122653,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 19209
    },
    {
      "epoch": 7.674790251697963,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 19210
    },
    {
      "epoch": 7.6751897722732725,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 19211
    },
    {
      "epoch": 7.675589292848581,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 19212
    },
    {
      "epoch": 7.675988813423891,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 19213
    },
    {
      "epoch": 7.676388333999201,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 19214
    },
    {
      "epoch": 7.676787854574511,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 19215
    },
    {
      "epoch": 7.6771873751498205,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 19216
    },
    {
      "epoch": 7.677586895725129,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 19217
    },
    {
      "epoch": 7.677986416300439,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 19218
    },
    {
      "epoch": 7.678385936875749,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 19219
    },
    {
      "epoch": 7.678785457451059,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 19220
    },
    {
      "epoch": 7.679184978026369,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 19221
    },
    {
      "epoch": 7.679584498601678,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 19222
    },
    {
      "epoch": 7.679984019176987,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 19223
    },
    {
      "epoch": 7.680383539752297,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 19224
    },
    {
      "epoch": 7.680783060327607,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 19225
    },
    {
      "epoch": 7.681182580902917,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 19226
    },
    {
      "epoch": 7.681582101478226,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 19227
    },
    {
      "epoch": 7.681981622053536,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 19228
    },
    {
      "epoch": 7.682381142628845,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 19229
    },
    {
      "epoch": 7.682780663204155,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 19230
    },
    {
      "epoch": 7.683180183779465,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 19231
    },
    {
      "epoch": 7.683579704354774,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 19232
    },
    {
      "epoch": 7.683979224930084,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 19233
    },
    {
      "epoch": 7.684378745505393,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 19234
    },
    {
      "epoch": 7.684778266080703,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 19235
    },
    {
      "epoch": 7.685177786656013,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 19236
    },
    {
      "epoch": 7.6855773072313225,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 19237
    },
    {
      "epoch": 7.685976827806632,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 19238
    },
    {
      "epoch": 7.686376348381942,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 19239
    },
    {
      "epoch": 7.686775868957251,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 19240
    },
    {
      "epoch": 7.687175389532561,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 19241
    },
    {
      "epoch": 7.6875749101078705,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 19242
    },
    {
      "epoch": 7.68797443068318,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 19243
    },
    {
      "epoch": 7.68837395125849,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 19244
    },
    {
      "epoch": 7.6887734718338,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 19245
    },
    {
      "epoch": 7.689172992409109,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 19246
    },
    {
      "epoch": 7.689572512984419,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0538,
      "step": 19247
    },
    {
      "epoch": 7.689972033559728,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 19248
    },
    {
      "epoch": 7.690371554135038,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 19249
    },
    {
      "epoch": 7.690771074710348,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 19250
    },
    {
      "epoch": 7.691170595285657,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 19251
    },
    {
      "epoch": 7.691570115860967,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 19252
    },
    {
      "epoch": 7.691969636436276,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 19253
    },
    {
      "epoch": 7.692369157011586,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 19254
    },
    {
      "epoch": 7.692768677586896,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 19255
    },
    {
      "epoch": 7.693168198162205,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 19256
    },
    {
      "epoch": 7.693567718737515,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 19257
    },
    {
      "epoch": 7.6939672393128244,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 19258
    },
    {
      "epoch": 7.694366759888134,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 19259
    },
    {
      "epoch": 7.694766280463444,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 19260
    },
    {
      "epoch": 7.695165801038754,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 19261
    },
    {
      "epoch": 7.695565321614064,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 19262
    },
    {
      "epoch": 7.6959648421893725,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 19263
    },
    {
      "epoch": 7.696364362764682,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 19264
    },
    {
      "epoch": 7.696763883339992,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 19265
    },
    {
      "epoch": 7.697163403915302,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 19266
    },
    {
      "epoch": 7.697562924490612,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 19267
    },
    {
      "epoch": 7.6979624450659205,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 19268
    },
    {
      "epoch": 7.69836196564123,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 19269
    },
    {
      "epoch": 7.69876148621654,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 19270
    },
    {
      "epoch": 7.69916100679185,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 19271
    },
    {
      "epoch": 7.69956052736716,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 19272
    },
    {
      "epoch": 7.699960047942469,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 19273
    },
    {
      "epoch": 7.700359568517778,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 19274
    },
    {
      "epoch": 7.700759089093088,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 19275
    },
    {
      "epoch": 7.701158609668398,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 19276
    },
    {
      "epoch": 7.701558130243708,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 19277
    },
    {
      "epoch": 7.7019576508190175,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 19278
    },
    {
      "epoch": 7.702357171394327,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 19279
    },
    {
      "epoch": 7.702756691969636,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 19280
    },
    {
      "epoch": 7.703156212544946,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 19281
    },
    {
      "epoch": 7.703555733120256,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 19282
    },
    {
      "epoch": 7.7039552536955656,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 19283
    },
    {
      "epoch": 7.704354774270875,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 19284
    },
    {
      "epoch": 7.704754294846184,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 19285
    },
    {
      "epoch": 7.705153815421494,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 19286
    },
    {
      "epoch": 7.705553335996804,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 19287
    },
    {
      "epoch": 7.705952856572114,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 19288
    },
    {
      "epoch": 7.706352377147423,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 19289
    },
    {
      "epoch": 7.706751897722732,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 19290
    },
    {
      "epoch": 7.707151418298042,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 19291
    },
    {
      "epoch": 7.707550938873352,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 19292
    },
    {
      "epoch": 7.707950459448662,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 19293
    },
    {
      "epoch": 7.708349980023971,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 19294
    },
    {
      "epoch": 7.708749500599281,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 19295
    },
    {
      "epoch": 7.70914902117459,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 19296
    },
    {
      "epoch": 7.7095485417499,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 19297
    },
    {
      "epoch": 7.70994806232521,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 19298
    },
    {
      "epoch": 7.7103475829005195,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 19299
    },
    {
      "epoch": 7.710747103475829,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 19300
    },
    {
      "epoch": 7.711146624051139,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 19301
    },
    {
      "epoch": 7.711546144626448,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 19302
    },
    {
      "epoch": 7.711945665201758,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 19303
    },
    {
      "epoch": 7.7123451857770675,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 19304
    },
    {
      "epoch": 7.712744706352377,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 19305
    },
    {
      "epoch": 7.713144226927687,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 19306
    },
    {
      "epoch": 7.713543747502996,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 19307
    },
    {
      "epoch": 7.713943268078306,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 19308
    },
    {
      "epoch": 7.714342788653616,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 19309
    },
    {
      "epoch": 7.714742309228925,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 19310
    },
    {
      "epoch": 7.715141829804235,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 19311
    },
    {
      "epoch": 7.715541350379545,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 19312
    },
    {
      "epoch": 7.715940870954854,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 19313
    },
    {
      "epoch": 7.716340391530164,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 19314
    },
    {
      "epoch": 7.716739912105473,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 19315
    },
    {
      "epoch": 7.717139432680783,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 19316
    },
    {
      "epoch": 7.717538953256093,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 19317
    },
    {
      "epoch": 7.717938473831403,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 19318
    },
    {
      "epoch": 7.718337994406712,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 19319
    },
    {
      "epoch": 7.718737514982021,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 19320
    },
    {
      "epoch": 7.719137035557331,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 19321
    },
    {
      "epoch": 7.719536556132641,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 19322
    },
    {
      "epoch": 7.719936076707951,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 19323
    },
    {
      "epoch": 7.72033559728326,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 19324
    },
    {
      "epoch": 7.7207351178585695,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 19325
    },
    {
      "epoch": 7.721134638433879,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 19326
    },
    {
      "epoch": 7.721534159009189,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 19327
    },
    {
      "epoch": 7.721933679584499,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 19328
    },
    {
      "epoch": 7.722333200159808,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 19329
    },
    {
      "epoch": 7.7227327207351175,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 19330
    },
    {
      "epoch": 7.723132241310427,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 19331
    },
    {
      "epoch": 7.723531761885737,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 19332
    },
    {
      "epoch": 7.723931282461047,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 19333
    },
    {
      "epoch": 7.724330803036357,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 19334
    },
    {
      "epoch": 7.7247303236116664,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 19335
    },
    {
      "epoch": 7.725129844186975,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 19336
    },
    {
      "epoch": 7.725529364762285,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 19337
    },
    {
      "epoch": 7.725928885337595,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 19338
    },
    {
      "epoch": 7.726328405912905,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 19339
    },
    {
      "epoch": 7.7267279264882145,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 19340
    },
    {
      "epoch": 7.727127447063523,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 19341
    },
    {
      "epoch": 7.727526967638833,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 19342
    },
    {
      "epoch": 7.727926488214143,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 19343
    },
    {
      "epoch": 7.728326008789453,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 19344
    },
    {
      "epoch": 7.7287255293647625,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 19345
    },
    {
      "epoch": 7.729125049940071,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 19346
    },
    {
      "epoch": 7.729524570515381,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 19347
    },
    {
      "epoch": 7.729924091090691,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 19348
    },
    {
      "epoch": 7.730323611666001,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 19349
    },
    {
      "epoch": 7.730723132241311,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 19350
    },
    {
      "epoch": 7.73112265281662,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 19351
    },
    {
      "epoch": 7.73152217339193,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 19352
    },
    {
      "epoch": 7.731921693967239,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 19353
    },
    {
      "epoch": 7.732321214542549,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 19354
    },
    {
      "epoch": 7.732720735117859,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 19355
    },
    {
      "epoch": 7.733120255693168,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 19356
    },
    {
      "epoch": 7.733519776268478,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 19357
    },
    {
      "epoch": 7.733919296843787,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0536,
      "step": 19358
    },
    {
      "epoch": 7.734318817419097,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 19359
    },
    {
      "epoch": 7.734718337994407,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 19360
    },
    {
      "epoch": 7.7351178585697165,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 19361
    },
    {
      "epoch": 7.735517379145026,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 19362
    },
    {
      "epoch": 7.735916899720335,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 19363
    },
    {
      "epoch": 7.736316420295645,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 19364
    },
    {
      "epoch": 7.736715940870955,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 19365
    },
    {
      "epoch": 7.7371154614462645,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0537,
      "step": 19366
    },
    {
      "epoch": 7.737514982021574,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 19367
    },
    {
      "epoch": 7.737914502596884,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 19368
    },
    {
      "epoch": 7.738314023172194,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 19369
    },
    {
      "epoch": 7.738713543747503,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 19370
    },
    {
      "epoch": 7.7391130643228125,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 19371
    },
    {
      "epoch": 7.739512584898122,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 19372
    },
    {
      "epoch": 7.739912105473432,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 19373
    },
    {
      "epoch": 7.740311626048742,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 19374
    },
    {
      "epoch": 7.740711146624051,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 19375
    },
    {
      "epoch": 7.741110667199361,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 19376
    },
    {
      "epoch": 7.74151018777467,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 19377
    },
    {
      "epoch": 7.74190970834998,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 19378
    },
    {
      "epoch": 7.74230922892529,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 19379
    },
    {
      "epoch": 7.742708749500599,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 19380
    },
    {
      "epoch": 7.743108270075909,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 19381
    },
    {
      "epoch": 7.743507790651218,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 19382
    },
    {
      "epoch": 7.743907311226528,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 19383
    },
    {
      "epoch": 7.744306831801838,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 19384
    },
    {
      "epoch": 7.744706352377148,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 19385
    },
    {
      "epoch": 7.745105872952457,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 19386
    },
    {
      "epoch": 7.7455053935277665,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 19387
    },
    {
      "epoch": 7.745904914103076,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 19388
    },
    {
      "epoch": 7.746304434678386,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 19389
    },
    {
      "epoch": 7.746703955253696,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 19390
    },
    {
      "epoch": 7.747103475829006,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 19391
    },
    {
      "epoch": 7.7475029964043145,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 19392
    },
    {
      "epoch": 7.747902516979624,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 19393
    },
    {
      "epoch": 7.748302037554934,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 19394
    },
    {
      "epoch": 7.748701558130244,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 19395
    },
    {
      "epoch": 7.749101078705554,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 19396
    },
    {
      "epoch": 7.7495005992808625,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 19397
    },
    {
      "epoch": 7.749900119856172,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 19398
    },
    {
      "epoch": 7.750299640431482,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 19399
    },
    {
      "epoch": 7.750699161006792,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 19400
    },
    {
      "epoch": 7.751098681582102,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 19401
    },
    {
      "epoch": 7.7514982021574115,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 19402
    },
    {
      "epoch": 7.75189772273272,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 19403
    },
    {
      "epoch": 7.75229724330803,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 19404
    },
    {
      "epoch": 7.75269676388334,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 19405
    },
    {
      "epoch": 7.75309628445865,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 19406
    },
    {
      "epoch": 7.7534958050339595,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 19407
    },
    {
      "epoch": 7.753895325609269,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 19408
    },
    {
      "epoch": 7.754294846184578,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 19409
    },
    {
      "epoch": 7.754694366759888,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 19410
    },
    {
      "epoch": 7.755093887335198,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 19411
    },
    {
      "epoch": 7.755493407910508,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 19412
    },
    {
      "epoch": 7.755892928485817,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 19413
    },
    {
      "epoch": 7.756292449061126,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 19414
    },
    {
      "epoch": 7.756691969636436,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 19415
    },
    {
      "epoch": 7.757091490211746,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 19416
    },
    {
      "epoch": 7.757491010787056,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 19417
    },
    {
      "epoch": 7.757890531362365,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 19418
    },
    {
      "epoch": 7.758290051937674,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 19419
    },
    {
      "epoch": 7.758689572512984,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 19420
    },
    {
      "epoch": 7.759089093088294,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 19421
    },
    {
      "epoch": 7.759488613663604,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 19422
    },
    {
      "epoch": 7.759888134238913,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 19423
    },
    {
      "epoch": 7.760287654814223,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 19424
    },
    {
      "epoch": 7.760687175389533,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 19425
    },
    {
      "epoch": 7.761086695964842,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 19426
    },
    {
      "epoch": 7.761486216540152,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 19427
    },
    {
      "epoch": 7.7618857371154615,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 19428
    },
    {
      "epoch": 7.762285257690771,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 19429
    },
    {
      "epoch": 7.762684778266081,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 19430
    },
    {
      "epoch": 7.76308429884139,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 19431
    },
    {
      "epoch": 7.7634838194167,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 19432
    },
    {
      "epoch": 7.7638833399920095,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 19433
    },
    {
      "epoch": 7.764282860567319,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 19434
    },
    {
      "epoch": 7.764682381142629,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 19435
    },
    {
      "epoch": 7.765081901717938,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 19436
    },
    {
      "epoch": 7.765481422293248,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 19437
    },
    {
      "epoch": 7.765880942868558,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 19438
    },
    {
      "epoch": 7.766280463443867,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 19439
    },
    {
      "epoch": 7.766679984019177,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 19440
    },
    {
      "epoch": 7.767079504594487,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 19441
    },
    {
      "epoch": 7.767479025169797,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 19442
    },
    {
      "epoch": 7.767878545745106,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 19443
    },
    {
      "epoch": 7.768278066320415,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 19444
    },
    {
      "epoch": 7.768677586895725,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 19445
    },
    {
      "epoch": 7.769077107471035,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 19446
    },
    {
      "epoch": 7.769476628046345,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 19447
    },
    {
      "epoch": 7.769876148621654,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 19448
    },
    {
      "epoch": 7.770275669196963,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 19449
    },
    {
      "epoch": 7.770675189772273,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 19450
    },
    {
      "epoch": 7.771074710347583,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 19451
    },
    {
      "epoch": 7.771474230922893,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 19452
    },
    {
      "epoch": 7.771873751498202,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 19453
    },
    {
      "epoch": 7.7722732720735115,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 19454
    },
    {
      "epoch": 7.772672792648821,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 19455
    },
    {
      "epoch": 7.773072313224131,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 19456
    },
    {
      "epoch": 7.773471833799441,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 19457
    },
    {
      "epoch": 7.773871354374751,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 19458
    },
    {
      "epoch": 7.77427087495006,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 19459
    },
    {
      "epoch": 7.774670395525369,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 19460
    },
    {
      "epoch": 7.775069916100679,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 19461
    },
    {
      "epoch": 7.775469436675989,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 19462
    },
    {
      "epoch": 7.775868957251299,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 19463
    },
    {
      "epoch": 7.7762684778266085,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 19464
    },
    {
      "epoch": 7.776667998401917,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 19465
    },
    {
      "epoch": 7.777067518977227,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 19466
    },
    {
      "epoch": 7.777467039552537,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 19467
    },
    {
      "epoch": 7.777866560127847,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 19468
    },
    {
      "epoch": 7.7782660807031565,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 19469
    },
    {
      "epoch": 7.778665601278465,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 19470
    },
    {
      "epoch": 7.779065121853775,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 19471
    },
    {
      "epoch": 7.779464642429085,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 19472
    },
    {
      "epoch": 7.779864163004395,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 19473
    },
    {
      "epoch": 7.7802636835797045,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 19474
    },
    {
      "epoch": 7.780663204155014,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 19475
    },
    {
      "epoch": 7.781062724730323,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 19476
    },
    {
      "epoch": 7.781462245305633,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 19477
    },
    {
      "epoch": 7.781861765880943,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 19478
    },
    {
      "epoch": 7.782261286456253,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 19479
    },
    {
      "epoch": 7.782660807031562,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 19480
    },
    {
      "epoch": 7.783060327606872,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 19481
    },
    {
      "epoch": 7.783459848182181,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 19482
    },
    {
      "epoch": 7.783859368757491,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 19483
    },
    {
      "epoch": 7.784258889332801,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 19484
    },
    {
      "epoch": 7.78465840990811,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 19485
    },
    {
      "epoch": 7.78505793048342,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 19486
    },
    {
      "epoch": 7.785457451058729,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 19487
    },
    {
      "epoch": 7.785856971634039,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 19488
    },
    {
      "epoch": 7.786256492209349,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 19489
    },
    {
      "epoch": 7.7866560127846585,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 19490
    },
    {
      "epoch": 7.787055533359968,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 19491
    },
    {
      "epoch": 7.787455053935278,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 19492
    },
    {
      "epoch": 7.787854574510587,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 19493
    },
    {
      "epoch": 7.788254095085897,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 19494
    },
    {
      "epoch": 7.7886536156612065,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 19495
    },
    {
      "epoch": 7.789053136236516,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 19496
    },
    {
      "epoch": 7.789452656811826,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 19497
    },
    {
      "epoch": 7.789852177387136,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 19498
    },
    {
      "epoch": 7.790251697962445,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 19499
    },
    {
      "epoch": 7.7906512185377546,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 19500
    },
    {
      "epoch": 7.791050739113064,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 19501
    },
    {
      "epoch": 7.791450259688374,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 19502
    },
    {
      "epoch": 7.791849780263684,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 19503
    },
    {
      "epoch": 7.792249300838993,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 19504
    },
    {
      "epoch": 7.792648821414303,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 19505
    },
    {
      "epoch": 7.793048341989612,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 19506
    },
    {
      "epoch": 7.793447862564922,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 19507
    },
    {
      "epoch": 7.793847383140232,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 19508
    },
    {
      "epoch": 7.794246903715541,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 19509
    },
    {
      "epoch": 7.794646424290851,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 19510
    },
    {
      "epoch": 7.79504594486616,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 19511
    },
    {
      "epoch": 7.79544546544147,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 19512
    },
    {
      "epoch": 7.79584498601678,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 19513
    },
    {
      "epoch": 7.79624450659209,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 19514
    },
    {
      "epoch": 7.7966440271674,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 19515
    },
    {
      "epoch": 7.7970435477427085,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 19516
    },
    {
      "epoch": 7.797443068318018,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 19517
    },
    {
      "epoch": 7.797842588893328,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 19518
    },
    {
      "epoch": 7.798242109468638,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 19519
    },
    {
      "epoch": 7.798641630043948,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 19520
    },
    {
      "epoch": 7.7990411506192565,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 19521
    },
    {
      "epoch": 7.799440671194566,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 19522
    },
    {
      "epoch": 7.799840191769876,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 19523
    },
    {
      "epoch": 7.800239712345186,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 19524
    },
    {
      "epoch": 7.800639232920496,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 19525
    },
    {
      "epoch": 7.801038753495805,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 19526
    },
    {
      "epoch": 7.801438274071114,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 19527
    },
    {
      "epoch": 7.801837794646424,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 19528
    },
    {
      "epoch": 7.802237315221734,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 19529
    },
    {
      "epoch": 7.802636835797044,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 19530
    },
    {
      "epoch": 7.8030363563723535,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 19531
    },
    {
      "epoch": 7.803435876947663,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 19532
    },
    {
      "epoch": 7.803835397522972,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 19533
    },
    {
      "epoch": 7.804234918098282,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 19534
    },
    {
      "epoch": 7.804634438673592,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 19535
    },
    {
      "epoch": 7.8050339592489015,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 19536
    },
    {
      "epoch": 7.805433479824211,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 19537
    },
    {
      "epoch": 7.80583300039952,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 19538
    },
    {
      "epoch": 7.80623252097483,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 19539
    },
    {
      "epoch": 7.80663204155014,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 19540
    },
    {
      "epoch": 7.80703156212545,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 19541
    },
    {
      "epoch": 7.807431082700759,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 19542
    },
    {
      "epoch": 7.807830603276068,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 19543
    },
    {
      "epoch": 7.808230123851378,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 19544
    },
    {
      "epoch": 7.808629644426688,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 19545
    },
    {
      "epoch": 7.809029165001998,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 19546
    },
    {
      "epoch": 7.809428685577307,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 19547
    },
    {
      "epoch": 7.809828206152617,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 19548
    },
    {
      "epoch": 7.810227726727927,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 19549
    },
    {
      "epoch": 7.810627247303236,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 19550
    },
    {
      "epoch": 7.811026767878546,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 19551
    },
    {
      "epoch": 7.8114262884538554,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 19552
    },
    {
      "epoch": 7.811825809029165,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 19553
    },
    {
      "epoch": 7.812225329604475,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 19554
    },
    {
      "epoch": 7.812624850179784,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 19555
    },
    {
      "epoch": 7.813024370755094,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 19556
    },
    {
      "epoch": 7.8134238913304035,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 19557
    },
    {
      "epoch": 7.813823411905713,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 19558
    },
    {
      "epoch": 7.814222932481023,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 19559
    },
    {
      "epoch": 7.814622453056332,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 19560
    },
    {
      "epoch": 7.815021973631642,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 19561
    },
    {
      "epoch": 7.8154214942069515,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 19562
    },
    {
      "epoch": 7.815821014782261,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 19563
    },
    {
      "epoch": 7.816220535357571,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 19564
    },
    {
      "epoch": 7.816620055932881,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 19565
    },
    {
      "epoch": 7.81701957650819,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 19566
    },
    {
      "epoch": 7.8174190970835,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 19567
    },
    {
      "epoch": 7.817818617658809,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 19568
    },
    {
      "epoch": 7.818218138234119,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 19569
    },
    {
      "epoch": 7.818617658809429,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 19570
    },
    {
      "epoch": 7.819017179384739,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 19571
    },
    {
      "epoch": 7.819416699960048,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 19572
    },
    {
      "epoch": 7.819816220535357,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 19573
    },
    {
      "epoch": 7.820215741110667,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 19574
    },
    {
      "epoch": 7.820615261685977,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 19575
    },
    {
      "epoch": 7.821014782261287,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 19576
    },
    {
      "epoch": 7.821414302836596,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 19577
    },
    {
      "epoch": 7.8218138234119055,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 19578
    },
    {
      "epoch": 7.822213343987215,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 19579
    },
    {
      "epoch": 7.822612864562525,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 19580
    },
    {
      "epoch": 7.823012385137835,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 19581
    },
    {
      "epoch": 7.823411905713145,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 19582
    },
    {
      "epoch": 7.8238114262884535,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 19583
    },
    {
      "epoch": 7.824210946863763,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 19584
    },
    {
      "epoch": 7.824610467439073,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 19585
    },
    {
      "epoch": 7.825009988014383,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 19586
    },
    {
      "epoch": 7.825409508589693,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 19587
    },
    {
      "epoch": 7.825809029165002,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 19588
    },
    {
      "epoch": 7.826208549740311,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 19589
    },
    {
      "epoch": 7.826608070315621,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 19590
    },
    {
      "epoch": 7.827007590890931,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 19591
    },
    {
      "epoch": 7.827407111466241,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 19592
    },
    {
      "epoch": 7.8278066320415505,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 19593
    },
    {
      "epoch": 7.828206152616859,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 19594
    },
    {
      "epoch": 7.828605673192169,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 19595
    },
    {
      "epoch": 7.829005193767479,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 19596
    },
    {
      "epoch": 7.829404714342789,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 19597
    },
    {
      "epoch": 7.8298042349180985,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 19598
    },
    {
      "epoch": 7.830203755493407,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 19599
    },
    {
      "epoch": 7.830603276068717,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 19600
    },
    {
      "epoch": 7.831002796644027,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 19601
    },
    {
      "epoch": 7.831402317219337,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 19602
    },
    {
      "epoch": 7.831801837794647,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 19603
    },
    {
      "epoch": 7.832201358369956,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 19604
    },
    {
      "epoch": 7.832600878945266,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 19605
    },
    {
      "epoch": 7.833000399520575,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 19606
    },
    {
      "epoch": 7.833399920095885,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 19607
    },
    {
      "epoch": 7.833799440671195,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 19608
    },
    {
      "epoch": 7.834198961246504,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 19609
    },
    {
      "epoch": 7.834598481821814,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 19610
    },
    {
      "epoch": 7.834998002397123,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 19611
    },
    {
      "epoch": 7.835397522972433,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 19612
    },
    {
      "epoch": 7.835797043547743,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 19613
    },
    {
      "epoch": 7.836196564123052,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 19614
    },
    {
      "epoch": 7.836596084698362,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 19615
    },
    {
      "epoch": 7.836995605273671,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 19616
    },
    {
      "epoch": 7.837395125848981,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 19617
    },
    {
      "epoch": 7.837794646424291,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 19618
    },
    {
      "epoch": 7.8381941669996005,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 19619
    },
    {
      "epoch": 7.83859368757491,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 19620
    },
    {
      "epoch": 7.83899320815022,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 19621
    },
    {
      "epoch": 7.83939272872553,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 19622
    },
    {
      "epoch": 7.839792249300839,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 19623
    },
    {
      "epoch": 7.8401917698761485,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 19624
    },
    {
      "epoch": 7.840591290451458,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 19625
    },
    {
      "epoch": 7.840990811026768,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 19626
    },
    {
      "epoch": 7.841390331602078,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 19627
    },
    {
      "epoch": 7.841789852177387,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 19628
    },
    {
      "epoch": 7.842189372752697,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 19629
    },
    {
      "epoch": 7.842588893328006,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 19630
    },
    {
      "epoch": 7.842988413903316,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 19631
    },
    {
      "epoch": 7.843387934478626,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 19632
    },
    {
      "epoch": 7.843787455053935,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 19633
    },
    {
      "epoch": 7.844186975629245,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 19634
    },
    {
      "epoch": 7.844586496204554,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 19635
    },
    {
      "epoch": 7.844986016779864,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 19636
    },
    {
      "epoch": 7.845385537355174,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 19637
    },
    {
      "epoch": 7.845785057930484,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 19638
    },
    {
      "epoch": 7.8461845785057935,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 19639
    },
    {
      "epoch": 7.846584099081102,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 19640
    },
    {
      "epoch": 7.846983619656412,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 19641
    },
    {
      "epoch": 7.847383140231722,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 19642
    },
    {
      "epoch": 7.847782660807032,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 19643
    },
    {
      "epoch": 7.848182181382342,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 19644
    },
    {
      "epoch": 7.8485817019576505,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 19645
    },
    {
      "epoch": 7.84898122253296,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 19646
    },
    {
      "epoch": 7.84938074310827,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 19647
    },
    {
      "epoch": 7.84978026368358,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 19648
    },
    {
      "epoch": 7.85017978425889,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 19649
    },
    {
      "epoch": 7.8505793048341985,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 19650
    },
    {
      "epoch": 7.850978825409508,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 19651
    },
    {
      "epoch": 7.851378345984818,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 19652
    },
    {
      "epoch": 7.851777866560128,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 19653
    },
    {
      "epoch": 7.852177387135438,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 19654
    },
    {
      "epoch": 7.8525769077107475,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 19655
    },
    {
      "epoch": 7.852976428286056,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 19656
    },
    {
      "epoch": 7.853375948861366,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 19657
    },
    {
      "epoch": 7.853775469436676,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 19658
    },
    {
      "epoch": 7.854174990011986,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 19659
    },
    {
      "epoch": 7.8545745105872955,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 19660
    },
    {
      "epoch": 7.854974031162605,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 19661
    },
    {
      "epoch": 7.855373551737914,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 19662
    },
    {
      "epoch": 7.855773072313224,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 19663
    },
    {
      "epoch": 7.856172592888534,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 19664
    },
    {
      "epoch": 7.8565721134638435,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 19665
    },
    {
      "epoch": 7.856971634039153,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 19666
    },
    {
      "epoch": 7.857371154614462,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 19667
    },
    {
      "epoch": 7.857770675189772,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 19668
    },
    {
      "epoch": 7.858170195765082,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 19669
    },
    {
      "epoch": 7.858569716340392,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 19670
    },
    {
      "epoch": 7.858969236915701,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 19671
    },
    {
      "epoch": 7.859368757491011,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 19672
    },
    {
      "epoch": 7.85976827806632,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 19673
    },
    {
      "epoch": 7.86016779864163,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 19674
    },
    {
      "epoch": 7.86056731921694,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 19675
    },
    {
      "epoch": 7.860966839792249,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 19676
    },
    {
      "epoch": 7.861366360367559,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 19677
    },
    {
      "epoch": 7.861765880942869,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 19678
    },
    {
      "epoch": 7.862165401518178,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 19679
    },
    {
      "epoch": 7.862564922093488,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 19680
    },
    {
      "epoch": 7.8629644426687975,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 19681
    },
    {
      "epoch": 7.863363963244107,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.057,
      "step": 19682
    },
    {
      "epoch": 7.863763483819417,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 19683
    },
    {
      "epoch": 7.864163004394726,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 19684
    },
    {
      "epoch": 7.864562524970036,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 19685
    },
    {
      "epoch": 7.8649620455453455,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 19686
    },
    {
      "epoch": 7.865361566120655,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 19687
    },
    {
      "epoch": 7.865761086695965,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 19688
    },
    {
      "epoch": 7.866160607271274,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 19689
    },
    {
      "epoch": 7.866560127846584,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 19690
    },
    {
      "epoch": 7.8669596484218935,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 19691
    },
    {
      "epoch": 7.867359168997203,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 19692
    },
    {
      "epoch": 7.867758689572513,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 19693
    },
    {
      "epoch": 7.868158210147823,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 19694
    },
    {
      "epoch": 7.868557730723133,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 19695
    },
    {
      "epoch": 7.868957251298442,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 19696
    },
    {
      "epoch": 7.869356771873751,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 19697
    },
    {
      "epoch": 7.869756292449061,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 19698
    },
    {
      "epoch": 7.870155813024371,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 19699
    },
    {
      "epoch": 7.870555333599681,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 19700
    },
    {
      "epoch": 7.87095485417499,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 19701
    },
    {
      "epoch": 7.871354374750299,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 19702
    },
    {
      "epoch": 7.871753895325609,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 19703
    },
    {
      "epoch": 7.872153415900919,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 19704
    },
    {
      "epoch": 7.872552936476229,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 19705
    },
    {
      "epoch": 7.872952457051538,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 19706
    },
    {
      "epoch": 7.8733519776268475,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 19707
    },
    {
      "epoch": 7.873751498202157,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 19708
    },
    {
      "epoch": 7.874151018777467,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 19709
    },
    {
      "epoch": 7.874550539352777,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 19710
    },
    {
      "epoch": 7.874950059928087,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 19711
    },
    {
      "epoch": 7.875349580503396,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 19712
    },
    {
      "epoch": 7.875749101078705,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 19713
    },
    {
      "epoch": 7.876148621654015,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 19714
    },
    {
      "epoch": 7.876548142229325,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 19715
    },
    {
      "epoch": 7.876947662804635,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 19716
    },
    {
      "epoch": 7.877347183379944,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 19717
    },
    {
      "epoch": 7.877746703955253,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 19718
    },
    {
      "epoch": 7.878146224530563,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 19719
    },
    {
      "epoch": 7.878545745105873,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 19720
    },
    {
      "epoch": 7.878945265681183,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 19721
    },
    {
      "epoch": 7.8793447862564925,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 19722
    },
    {
      "epoch": 7.879744306831801,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 19723
    },
    {
      "epoch": 7.880143827407111,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 19724
    },
    {
      "epoch": 7.880543347982421,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 19725
    },
    {
      "epoch": 7.880942868557731,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 19726
    },
    {
      "epoch": 7.8813423891330405,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 19727
    },
    {
      "epoch": 7.88174190970835,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 19728
    },
    {
      "epoch": 7.88214143028366,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 19729
    },
    {
      "epoch": 7.882540950858969,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 19730
    },
    {
      "epoch": 7.882940471434279,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 19731
    },
    {
      "epoch": 7.883339992009589,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 19732
    },
    {
      "epoch": 7.883739512584898,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 19733
    },
    {
      "epoch": 7.884139033160208,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 19734
    },
    {
      "epoch": 7.884538553735517,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 19735
    },
    {
      "epoch": 7.884938074310827,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 19736
    },
    {
      "epoch": 7.885337594886137,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 19737
    },
    {
      "epoch": 7.885737115461446,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 19738
    },
    {
      "epoch": 7.886136636036756,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 19739
    },
    {
      "epoch": 7.886536156612065,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 19740
    },
    {
      "epoch": 7.886935677187375,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 19741
    },
    {
      "epoch": 7.887335197762685,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 19742
    },
    {
      "epoch": 7.887734718337994,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 19743
    },
    {
      "epoch": 7.888134238913304,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 19744
    },
    {
      "epoch": 7.888533759488614,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 19745
    },
    {
      "epoch": 7.888933280063923,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 19746
    },
    {
      "epoch": 7.889332800639233,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 19747
    },
    {
      "epoch": 7.8897323212145425,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 19748
    },
    {
      "epoch": 7.890131841789852,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 19749
    },
    {
      "epoch": 7.890531362365162,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 19750
    },
    {
      "epoch": 7.890930882940472,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 19751
    },
    {
      "epoch": 7.891330403515781,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 19752
    },
    {
      "epoch": 7.8917299240910905,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 19753
    },
    {
      "epoch": 7.8921294446664,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 19754
    },
    {
      "epoch": 7.89252896524171,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 19755
    },
    {
      "epoch": 7.89292848581702,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 19756
    },
    {
      "epoch": 7.893328006392329,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 19757
    },
    {
      "epoch": 7.893727526967639,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 19758
    },
    {
      "epoch": 7.894127047542948,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 19759
    },
    {
      "epoch": 7.894526568118258,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 19760
    },
    {
      "epoch": 7.894926088693568,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 19761
    },
    {
      "epoch": 7.895325609268878,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 19762
    },
    {
      "epoch": 7.895725129844187,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 19763
    },
    {
      "epoch": 7.896124650419496,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 19764
    },
    {
      "epoch": 7.896524170994806,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 19765
    },
    {
      "epoch": 7.896923691570116,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 19766
    },
    {
      "epoch": 7.897323212145426,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 19767
    },
    {
      "epoch": 7.8977227327207356,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 19768
    },
    {
      "epoch": 7.8981222532960444,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 19769
    },
    {
      "epoch": 7.898521773871354,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 19770
    },
    {
      "epoch": 7.898921294446664,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 19771
    },
    {
      "epoch": 7.899320815021974,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 19772
    },
    {
      "epoch": 7.899720335597284,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 19773
    },
    {
      "epoch": 7.9001198561725925,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 19774
    },
    {
      "epoch": 7.900519376747902,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 19775
    },
    {
      "epoch": 7.900918897323212,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 19776
    },
    {
      "epoch": 7.901318417898522,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 19777
    },
    {
      "epoch": 7.901717938473832,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 19778
    },
    {
      "epoch": 7.9021174590491405,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 19779
    },
    {
      "epoch": 7.90251697962445,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 19780
    },
    {
      "epoch": 7.90291650019976,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 19781
    },
    {
      "epoch": 7.90331602077507,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 19782
    },
    {
      "epoch": 7.90371554135038,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 19783
    },
    {
      "epoch": 7.9041150619256895,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 19784
    },
    {
      "epoch": 7.904514582500999,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 19785
    },
    {
      "epoch": 7.904914103076308,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 19786
    },
    {
      "epoch": 7.905313623651618,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 19787
    },
    {
      "epoch": 7.905713144226928,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 19788
    },
    {
      "epoch": 7.9061126648022375,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 19789
    },
    {
      "epoch": 7.906512185377547,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 19790
    },
    {
      "epoch": 7.906911705952856,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 19791
    },
    {
      "epoch": 7.907311226528166,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 19792
    },
    {
      "epoch": 7.907710747103476,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 19793
    },
    {
      "epoch": 7.9081102676787856,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 19794
    },
    {
      "epoch": 7.908509788254095,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 19795
    },
    {
      "epoch": 7.908909308829404,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 19796
    },
    {
      "epoch": 7.909308829404714,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 19797
    },
    {
      "epoch": 7.909708349980024,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 19798
    },
    {
      "epoch": 7.910107870555334,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 19799
    },
    {
      "epoch": 7.910507391130643,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 19800
    },
    {
      "epoch": 7.910906911705953,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 19801
    },
    {
      "epoch": 7.911306432281263,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 19802
    },
    {
      "epoch": 7.911705952856572,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 19803
    },
    {
      "epoch": 7.912105473431882,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 19804
    },
    {
      "epoch": 7.912504994007191,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 19805
    },
    {
      "epoch": 7.912904514582501,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 19806
    },
    {
      "epoch": 7.913304035157811,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 19807
    },
    {
      "epoch": 7.91370355573312,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 19808
    },
    {
      "epoch": 7.91410307630843,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 19809
    },
    {
      "epoch": 7.9145025968837395,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 19810
    },
    {
      "epoch": 7.914902117459049,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 19811
    },
    {
      "epoch": 7.915301638034359,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 19812
    },
    {
      "epoch": 7.915701158609668,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 19813
    },
    {
      "epoch": 7.916100679184978,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 19814
    },
    {
      "epoch": 7.9165001997602875,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 19815
    },
    {
      "epoch": 7.916899720335597,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 19816
    },
    {
      "epoch": 7.917299240910907,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 19817
    },
    {
      "epoch": 7.917698761486217,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 19818
    },
    {
      "epoch": 7.918098282061527,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 19819
    },
    {
      "epoch": 7.918497802636836,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 19820
    },
    {
      "epoch": 7.918897323212145,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 19821
    },
    {
      "epoch": 7.919296843787455,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 19822
    },
    {
      "epoch": 7.919696364362765,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 19823
    },
    {
      "epoch": 7.920095884938075,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 19824
    },
    {
      "epoch": 7.920495405513384,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 19825
    },
    {
      "epoch": 7.920894926088693,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 19826
    },
    {
      "epoch": 7.921294446664003,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 19827
    },
    {
      "epoch": 7.921693967239313,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 19828
    },
    {
      "epoch": 7.922093487814623,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 19829
    },
    {
      "epoch": 7.922493008389932,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0529,
      "step": 19830
    },
    {
      "epoch": 7.922892528965241,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 19831
    },
    {
      "epoch": 7.923292049540551,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 19832
    },
    {
      "epoch": 7.923691570115861,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 19833
    },
    {
      "epoch": 7.924091090691171,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 19834
    },
    {
      "epoch": 7.924490611266481,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 19835
    },
    {
      "epoch": 7.9248901318417895,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 19836
    },
    {
      "epoch": 7.925289652417099,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 19837
    },
    {
      "epoch": 7.925689172992409,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 19838
    },
    {
      "epoch": 7.926088693567719,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 19839
    },
    {
      "epoch": 7.926488214143029,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 19840
    },
    {
      "epoch": 7.926887734718338,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 19841
    },
    {
      "epoch": 7.927287255293647,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 19842
    },
    {
      "epoch": 7.927686775868957,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 19843
    },
    {
      "epoch": 7.928086296444267,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 19844
    },
    {
      "epoch": 7.928485817019577,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 19845
    },
    {
      "epoch": 7.9288853375948865,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 19846
    },
    {
      "epoch": 7.929284858170195,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 19847
    },
    {
      "epoch": 7.929684378745505,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 19848
    },
    {
      "epoch": 7.930083899320815,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 19849
    },
    {
      "epoch": 7.930483419896125,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 19850
    },
    {
      "epoch": 7.9308829404714345,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 19851
    },
    {
      "epoch": 7.931282461046744,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 19852
    },
    {
      "epoch": 7.931681981622053,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 19853
    },
    {
      "epoch": 7.932081502197363,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 19854
    },
    {
      "epoch": 7.932481022772673,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 19855
    },
    {
      "epoch": 7.9328805433479825,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 19856
    },
    {
      "epoch": 7.933280063923292,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 19857
    },
    {
      "epoch": 7.933679584498602,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 19858
    },
    {
      "epoch": 7.934079105073911,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 19859
    },
    {
      "epoch": 7.934478625649221,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 19860
    },
    {
      "epoch": 7.934878146224531,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 19861
    },
    {
      "epoch": 7.93527766679984,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 19862
    },
    {
      "epoch": 7.93567718737515,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 19863
    },
    {
      "epoch": 7.936076707950459,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 19864
    },
    {
      "epoch": 7.936476228525769,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 19865
    },
    {
      "epoch": 7.936875749101079,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 19866
    },
    {
      "epoch": 7.937275269676388,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 19867
    },
    {
      "epoch": 7.937674790251698,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 19868
    },
    {
      "epoch": 7.938074310827007,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 19869
    },
    {
      "epoch": 7.938473831402317,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 19870
    },
    {
      "epoch": 7.938873351977627,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 19871
    },
    {
      "epoch": 7.9392728725529365,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 19872
    },
    {
      "epoch": 7.939672393128246,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 19873
    },
    {
      "epoch": 7.940071913703556,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 19874
    },
    {
      "epoch": 7.940471434278866,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 19875
    },
    {
      "epoch": 7.940870954854175,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 19876
    },
    {
      "epoch": 7.9412704754294845,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 19877
    },
    {
      "epoch": 7.941669996004794,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 19878
    },
    {
      "epoch": 7.942069516580104,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 19879
    },
    {
      "epoch": 7.942469037155414,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 19880
    },
    {
      "epoch": 7.942868557730723,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 19881
    },
    {
      "epoch": 7.9432680783060325,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 19882
    },
    {
      "epoch": 7.943667598881342,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 19883
    },
    {
      "epoch": 7.944067119456652,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 19884
    },
    {
      "epoch": 7.944466640031962,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 19885
    },
    {
      "epoch": 7.944866160607271,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 19886
    },
    {
      "epoch": 7.945265681182581,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 19887
    },
    {
      "epoch": 7.94566520175789,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 19888
    },
    {
      "epoch": 7.9460647223332,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 19889
    },
    {
      "epoch": 7.94646424290851,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 19890
    },
    {
      "epoch": 7.94686376348382,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 19891
    },
    {
      "epoch": 7.9472632840591295,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 19892
    },
    {
      "epoch": 7.947662804634438,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 19893
    },
    {
      "epoch": 7.948062325209748,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 19894
    },
    {
      "epoch": 7.948461845785058,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 19895
    },
    {
      "epoch": 7.948861366360368,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 19896
    },
    {
      "epoch": 7.949260886935678,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 19897
    },
    {
      "epoch": 7.9496604075109865,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 19898
    },
    {
      "epoch": 7.950059928086296,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 19899
    },
    {
      "epoch": 7.950459448661606,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 19900
    },
    {
      "epoch": 7.950858969236916,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 19901
    },
    {
      "epoch": 7.951258489812226,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 19902
    },
    {
      "epoch": 7.9516580103875345,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 19903
    },
    {
      "epoch": 7.952057530962844,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 19904
    },
    {
      "epoch": 7.952457051538154,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 19905
    },
    {
      "epoch": 7.952856572113464,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 19906
    },
    {
      "epoch": 7.953256092688774,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 19907
    },
    {
      "epoch": 7.953655613264083,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 19908
    },
    {
      "epoch": 7.954055133839393,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 19909
    },
    {
      "epoch": 7.954454654414702,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 19910
    },
    {
      "epoch": 7.954854174990012,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 19911
    },
    {
      "epoch": 7.955253695565322,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 19912
    },
    {
      "epoch": 7.9556532161406315,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 19913
    },
    {
      "epoch": 7.956052736715941,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 19914
    },
    {
      "epoch": 7.95645225729125,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 19915
    },
    {
      "epoch": 7.95685177786656,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 19916
    },
    {
      "epoch": 7.95725129844187,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 19917
    },
    {
      "epoch": 7.9576508190171795,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 19918
    },
    {
      "epoch": 7.958050339592489,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 19919
    },
    {
      "epoch": 7.958449860167798,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 19920
    },
    {
      "epoch": 7.958849380743108,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 19921
    },
    {
      "epoch": 7.959248901318418,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 19922
    },
    {
      "epoch": 7.959648421893728,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 19923
    },
    {
      "epoch": 7.960047942469037,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 19924
    },
    {
      "epoch": 7.960447463044347,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 19925
    },
    {
      "epoch": 7.960846983619656,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 19926
    },
    {
      "epoch": 7.961246504194966,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 19927
    },
    {
      "epoch": 7.961646024770276,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 19928
    },
    {
      "epoch": 7.962045545345585,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 19929
    },
    {
      "epoch": 7.962445065920895,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 19930
    },
    {
      "epoch": 7.962844586496205,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 19931
    },
    {
      "epoch": 7.963244107071514,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 19932
    },
    {
      "epoch": 7.963643627646824,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 19933
    },
    {
      "epoch": 7.964043148222133,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 19934
    },
    {
      "epoch": 7.964442668797443,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 19935
    },
    {
      "epoch": 7.964842189372753,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 19936
    },
    {
      "epoch": 7.965241709948062,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 19937
    },
    {
      "epoch": 7.965641230523372,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 19938
    },
    {
      "epoch": 7.9660407510986815,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 19939
    },
    {
      "epoch": 7.966440271673991,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 19940
    },
    {
      "epoch": 7.966839792249301,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 19941
    },
    {
      "epoch": 7.967239312824611,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 19942
    },
    {
      "epoch": 7.96763883339992,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 19943
    },
    {
      "epoch": 7.9680383539752295,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 19944
    },
    {
      "epoch": 7.968437874550539,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 19945
    },
    {
      "epoch": 7.968837395125849,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 19946
    },
    {
      "epoch": 7.969236915701159,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 19947
    },
    {
      "epoch": 7.969636436276469,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 19948
    },
    {
      "epoch": 7.970035956851778,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 19949
    },
    {
      "epoch": 7.970435477427087,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 19950
    },
    {
      "epoch": 7.970834998002397,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 19951
    },
    {
      "epoch": 7.971234518577707,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 19952
    },
    {
      "epoch": 7.971634039153017,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 19953
    },
    {
      "epoch": 7.972033559728326,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 19954
    },
    {
      "epoch": 7.972433080303635,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 19955
    },
    {
      "epoch": 7.972832600878945,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 19956
    },
    {
      "epoch": 7.973232121454255,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 19957
    },
    {
      "epoch": 7.973631642029565,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 19958
    },
    {
      "epoch": 7.974031162604874,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 19959
    },
    {
      "epoch": 7.974430683180183,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 19960
    },
    {
      "epoch": 7.974830203755493,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 19961
    },
    {
      "epoch": 7.975229724330803,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 19962
    },
    {
      "epoch": 7.975629244906113,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 19963
    },
    {
      "epoch": 7.976028765481423,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 19964
    },
    {
      "epoch": 7.976428286056732,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 19965
    },
    {
      "epoch": 7.976827806632041,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 19966
    },
    {
      "epoch": 7.977227327207351,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 19967
    },
    {
      "epoch": 7.977626847782661,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 19968
    },
    {
      "epoch": 7.978026368357971,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 19969
    },
    {
      "epoch": 7.97842588893328,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 19970
    },
    {
      "epoch": 7.978825409508589,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 19971
    },
    {
      "epoch": 7.979224930083899,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 19972
    },
    {
      "epoch": 7.979624450659209,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 19973
    },
    {
      "epoch": 7.980023971234519,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 19974
    },
    {
      "epoch": 7.9804234918098285,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 19975
    },
    {
      "epoch": 7.980823012385137,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 19976
    },
    {
      "epoch": 7.981222532960447,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 19977
    },
    {
      "epoch": 7.981622053535757,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 19978
    },
    {
      "epoch": 7.982021574111067,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 19979
    },
    {
      "epoch": 7.9824210946863765,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 19980
    },
    {
      "epoch": 7.982820615261686,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 19981
    },
    {
      "epoch": 7.983220135836996,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 19982
    },
    {
      "epoch": 7.983619656412305,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 19983
    },
    {
      "epoch": 7.984019176987615,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 19984
    },
    {
      "epoch": 7.9844186975629245,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 19985
    },
    {
      "epoch": 7.984818218138234,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 19986
    },
    {
      "epoch": 7.985217738713544,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 19987
    },
    {
      "epoch": 7.985617259288853,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 19988
    },
    {
      "epoch": 7.986016779864163,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 19989
    },
    {
      "epoch": 7.986416300439473,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 19990
    },
    {
      "epoch": 7.986815821014782,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 19991
    },
    {
      "epoch": 7.987215341590092,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 19992
    },
    {
      "epoch": 7.987614862165401,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 19993
    },
    {
      "epoch": 7.988014382740711,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 19994
    },
    {
      "epoch": 7.988413903316021,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 19995
    },
    {
      "epoch": 7.98881342389133,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 19996
    },
    {
      "epoch": 7.98921294446664,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 19997
    },
    {
      "epoch": 7.98961246504195,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 19998
    },
    {
      "epoch": 7.99001198561726,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 19999
    },
    {
      "epoch": 7.990411506192569,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 20000
    },
    {
      "epoch": 7.9908110267678785,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 20001
    },
    {
      "epoch": 7.991210547343188,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 20002
    },
    {
      "epoch": 7.991610067918498,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 20003
    },
    {
      "epoch": 7.992009588493808,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 20004
    },
    {
      "epoch": 7.992409109069117,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 20005
    },
    {
      "epoch": 7.9928086296444265,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 20006
    },
    {
      "epoch": 7.993208150219736,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 20007
    },
    {
      "epoch": 7.993607670795046,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 20008
    },
    {
      "epoch": 7.994007191370356,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 20009
    },
    {
      "epoch": 7.994406711945665,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 20010
    },
    {
      "epoch": 7.9948062325209746,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 20011
    },
    {
      "epoch": 7.995205753096284,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 20012
    },
    {
      "epoch": 7.995605273671594,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 20013
    },
    {
      "epoch": 7.996004794246904,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0521,
      "step": 20014
    },
    {
      "epoch": 7.996404314822214,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 20015
    },
    {
      "epoch": 7.996803835397523,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 20016
    },
    {
      "epoch": 7.997203355972832,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 20017
    },
    {
      "epoch": 7.997602876548142,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 20018
    },
    {
      "epoch": 7.998002397123452,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 20019
    },
    {
      "epoch": 7.998401917698762,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 20020
    },
    {
      "epoch": 7.9988014382740715,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 20021
    },
    {
      "epoch": 7.99920095884938,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 20022
    },
    {
      "epoch": 7.99960047942469,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 20023
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 20024
    },
    {
      "epoch": 8.000399520575309,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 20025
    },
    {
      "epoch": 8.00079904115062,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 20026
    },
    {
      "epoch": 8.001198561725928,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 20027
    },
    {
      "epoch": 8.00159808230124,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 20028
    },
    {
      "epoch": 8.001997602876548,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 20029
    },
    {
      "epoch": 8.002397123451857,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 20030
    },
    {
      "epoch": 8.002796644027168,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 20031
    },
    {
      "epoch": 8.003196164602477,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 20032
    },
    {
      "epoch": 8.003595685177787,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 20033
    },
    {
      "epoch": 8.003995205753096,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 20034
    },
    {
      "epoch": 8.004394726328407,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 20035
    },
    {
      "epoch": 8.004794246903716,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 20036
    },
    {
      "epoch": 8.005193767479025,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 20037
    },
    {
      "epoch": 8.005593288054335,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 20038
    },
    {
      "epoch": 8.005992808629644,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 20039
    },
    {
      "epoch": 8.006392329204955,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 20040
    },
    {
      "epoch": 8.006791849780264,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 20041
    },
    {
      "epoch": 8.007191370355573,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 20042
    },
    {
      "epoch": 8.007590890930883,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 20043
    },
    {
      "epoch": 8.007990411506192,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 20044
    },
    {
      "epoch": 8.008389932081503,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 20045
    },
    {
      "epoch": 8.008789452656812,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 20046
    },
    {
      "epoch": 8.00918897323212,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 20047
    },
    {
      "epoch": 8.009588493807431,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 20048
    },
    {
      "epoch": 8.00998801438274,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 20049
    },
    {
      "epoch": 8.010387534958051,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 20050
    },
    {
      "epoch": 8.01078705553336,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 20051
    },
    {
      "epoch": 8.01118657610867,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 20052
    },
    {
      "epoch": 8.01158609668398,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 20053
    },
    {
      "epoch": 8.011985617259288,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 20054
    },
    {
      "epoch": 8.012385137834599,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 20055
    },
    {
      "epoch": 8.012784658409908,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 20056
    },
    {
      "epoch": 8.013184178985219,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 20057
    },
    {
      "epoch": 8.013583699560527,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 20058
    },
    {
      "epoch": 8.013983220135836,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 20059
    },
    {
      "epoch": 8.014382740711147,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 20060
    },
    {
      "epoch": 8.014782261286456,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 20061
    },
    {
      "epoch": 8.015181781861767,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 20062
    },
    {
      "epoch": 8.015581302437075,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 20063
    },
    {
      "epoch": 8.015980823012384,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 20064
    },
    {
      "epoch": 8.016380343587695,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 20065
    },
    {
      "epoch": 8.016779864163004,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 20066
    },
    {
      "epoch": 8.017179384738315,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 20067
    },
    {
      "epoch": 8.017578905313623,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 20068
    },
    {
      "epoch": 8.017978425888934,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 20069
    },
    {
      "epoch": 8.018377946464243,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 20070
    },
    {
      "epoch": 8.018777467039552,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 20071
    },
    {
      "epoch": 8.019176987614863,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 20072
    },
    {
      "epoch": 8.019576508190172,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 20073
    },
    {
      "epoch": 8.019976028765482,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 20074
    },
    {
      "epoch": 8.020375549340791,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 20075
    },
    {
      "epoch": 8.0207750699161,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 20076
    },
    {
      "epoch": 8.02117459049141,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 20077
    },
    {
      "epoch": 8.02157411106672,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 20078
    },
    {
      "epoch": 8.02197363164203,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 20079
    },
    {
      "epoch": 8.02237315221734,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 20080
    },
    {
      "epoch": 8.022772672792648,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 20081
    },
    {
      "epoch": 8.023172193367959,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 20082
    },
    {
      "epoch": 8.023571713943268,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 20083
    },
    {
      "epoch": 8.023971234518578,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 20084
    },
    {
      "epoch": 8.024370755093887,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 20085
    },
    {
      "epoch": 8.024770275669196,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 20086
    },
    {
      "epoch": 8.025169796244507,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 20087
    },
    {
      "epoch": 8.025569316819816,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 20088
    },
    {
      "epoch": 8.025968837395126,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 20089
    },
    {
      "epoch": 8.026368357970435,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 20090
    },
    {
      "epoch": 8.026767878545746,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 20091
    },
    {
      "epoch": 8.027167399121055,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 20092
    },
    {
      "epoch": 8.027566919696364,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 20093
    },
    {
      "epoch": 8.027966440271674,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 20094
    },
    {
      "epoch": 8.028365960846983,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 20095
    },
    {
      "epoch": 8.028765481422294,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 20096
    },
    {
      "epoch": 8.029165001997603,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 20097
    },
    {
      "epoch": 8.029564522572912,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 20098
    },
    {
      "epoch": 8.029964043148222,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 20099
    },
    {
      "epoch": 8.030363563723531,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 20100
    },
    {
      "epoch": 8.030763084298842,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 20101
    },
    {
      "epoch": 8.031162604874151,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 20102
    },
    {
      "epoch": 8.03156212544946,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 20103
    },
    {
      "epoch": 8.03196164602477,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 20104
    },
    {
      "epoch": 8.03236116660008,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 20105
    },
    {
      "epoch": 8.03276068717539,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 20106
    },
    {
      "epoch": 8.033160207750699,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 20107
    },
    {
      "epoch": 8.03355972832601,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 20108
    },
    {
      "epoch": 8.033959248901319,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 20109
    },
    {
      "epoch": 8.034358769476627,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 20110
    },
    {
      "epoch": 8.034758290051938,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 20111
    },
    {
      "epoch": 8.035157810627247,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 20112
    },
    {
      "epoch": 8.035557331202558,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 20113
    },
    {
      "epoch": 8.035956851777867,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 20114
    },
    {
      "epoch": 8.036356372353175,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 20115
    },
    {
      "epoch": 8.036755892928486,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 20116
    },
    {
      "epoch": 8.037155413503795,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 20117
    },
    {
      "epoch": 8.037554934079106,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 20118
    },
    {
      "epoch": 8.037954454654415,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 20119
    },
    {
      "epoch": 8.038353975229724,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 20120
    },
    {
      "epoch": 8.038753495805034,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 20121
    },
    {
      "epoch": 8.039153016380343,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 20122
    },
    {
      "epoch": 8.039552536955654,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 20123
    },
    {
      "epoch": 8.039952057530963,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 20124
    },
    {
      "epoch": 8.040351578106273,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 20125
    },
    {
      "epoch": 8.040751098681582,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 20126
    },
    {
      "epoch": 8.041150619256891,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 20127
    },
    {
      "epoch": 8.041550139832202,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 20128
    },
    {
      "epoch": 8.04194966040751,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 20129
    },
    {
      "epoch": 8.042349180982821,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 20130
    },
    {
      "epoch": 8.04274870155813,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 20131
    },
    {
      "epoch": 8.04314822213344,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 20132
    },
    {
      "epoch": 8.04354774270875,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 20133
    },
    {
      "epoch": 8.043947263284059,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 20134
    },
    {
      "epoch": 8.04434678385937,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 20135
    },
    {
      "epoch": 8.044746304434678,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 20136
    },
    {
      "epoch": 8.045145825009987,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 20137
    },
    {
      "epoch": 8.045545345585298,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 20138
    },
    {
      "epoch": 8.045944866160607,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 20139
    },
    {
      "epoch": 8.046344386735917,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 20140
    },
    {
      "epoch": 8.046743907311226,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 20141
    },
    {
      "epoch": 8.047143427886537,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 20142
    },
    {
      "epoch": 8.047542948461846,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 20143
    },
    {
      "epoch": 8.047942469037155,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 20144
    },
    {
      "epoch": 8.048341989612465,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 20145
    },
    {
      "epoch": 8.048741510187774,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 20146
    },
    {
      "epoch": 8.049141030763085,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 20147
    },
    {
      "epoch": 8.049540551338394,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 20148
    },
    {
      "epoch": 8.049940071913703,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 20149
    },
    {
      "epoch": 8.050339592489014,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 20150
    },
    {
      "epoch": 8.050739113064322,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 20151
    },
    {
      "epoch": 8.051138633639633,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 20152
    },
    {
      "epoch": 8.051538154214942,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 20153
    },
    {
      "epoch": 8.051937674790251,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 20154
    },
    {
      "epoch": 8.052337195365562,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 20155
    },
    {
      "epoch": 8.05273671594087,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 20156
    },
    {
      "epoch": 8.053136236516181,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 20157
    },
    {
      "epoch": 8.05353575709149,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 20158
    },
    {
      "epoch": 8.053935277666799,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 20159
    },
    {
      "epoch": 8.05433479824211,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 20160
    },
    {
      "epoch": 8.054734318817419,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 20161
    },
    {
      "epoch": 8.05513383939273,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 20162
    },
    {
      "epoch": 8.055533359968038,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 20163
    },
    {
      "epoch": 8.055932880543349,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 20164
    },
    {
      "epoch": 8.056332401118658,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 20165
    },
    {
      "epoch": 8.056731921693967,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 20166
    },
    {
      "epoch": 8.057131442269277,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 20167
    },
    {
      "epoch": 8.057530962844586,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 20168
    },
    {
      "epoch": 8.057930483419897,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 20169
    },
    {
      "epoch": 8.058330003995206,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 20170
    },
    {
      "epoch": 8.058729524570515,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 20171
    },
    {
      "epoch": 8.059129045145825,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 20172
    },
    {
      "epoch": 8.059528565721134,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 20173
    },
    {
      "epoch": 8.059928086296445,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 20174
    },
    {
      "epoch": 8.060327606871754,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 20175
    },
    {
      "epoch": 8.060727127447063,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 20176
    },
    {
      "epoch": 8.061126648022373,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 20177
    },
    {
      "epoch": 8.061526168597682,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 20178
    },
    {
      "epoch": 8.061925689172993,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 20179
    },
    {
      "epoch": 8.062325209748302,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 20180
    },
    {
      "epoch": 8.062724730323612,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 20181
    },
    {
      "epoch": 8.063124250898921,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 20182
    },
    {
      "epoch": 8.06352377147423,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 20183
    },
    {
      "epoch": 8.063923292049541,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 20184
    },
    {
      "epoch": 8.06432281262485,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 20185
    },
    {
      "epoch": 8.06472233320016,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 20186
    },
    {
      "epoch": 8.06512185377547,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 20187
    },
    {
      "epoch": 8.065521374350778,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 20188
    },
    {
      "epoch": 8.065920894926089,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 20189
    },
    {
      "epoch": 8.066320415501398,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 20190
    },
    {
      "epoch": 8.066719936076709,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 20191
    },
    {
      "epoch": 8.067119456652017,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 20192
    },
    {
      "epoch": 8.067518977227326,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 20193
    },
    {
      "epoch": 8.067918497802637,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 20194
    },
    {
      "epoch": 8.068318018377946,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 20195
    },
    {
      "epoch": 8.068717538953257,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 20196
    },
    {
      "epoch": 8.069117059528566,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 20197
    },
    {
      "epoch": 8.069516580103876,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 20198
    },
    {
      "epoch": 8.069916100679185,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 20199
    },
    {
      "epoch": 8.070315621254494,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 20200
    },
    {
      "epoch": 8.070715141829805,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 20201
    },
    {
      "epoch": 8.071114662405114,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 20202
    },
    {
      "epoch": 8.071514182980424,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 20203
    },
    {
      "epoch": 8.071913703555733,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 20204
    },
    {
      "epoch": 8.072313224131042,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 20205
    },
    {
      "epoch": 8.072712744706353,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 20206
    },
    {
      "epoch": 8.073112265281662,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 20207
    },
    {
      "epoch": 8.073511785856972,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 20208
    },
    {
      "epoch": 8.073911306432281,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 20209
    },
    {
      "epoch": 8.07431082700759,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 20210
    },
    {
      "epoch": 8.0747103475829,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 20211
    },
    {
      "epoch": 8.07510986815821,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 20212
    },
    {
      "epoch": 8.07550938873352,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 20213
    },
    {
      "epoch": 8.07590890930883,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 20214
    },
    {
      "epoch": 8.07630842988414,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 20215
    },
    {
      "epoch": 8.076707950459449,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 20216
    },
    {
      "epoch": 8.077107471034758,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 20217
    },
    {
      "epoch": 8.077506991610068,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 20218
    },
    {
      "epoch": 8.077906512185377,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 20219
    },
    {
      "epoch": 8.078306032760688,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 20220
    },
    {
      "epoch": 8.078705553335997,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 20221
    },
    {
      "epoch": 8.079105073911306,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 20222
    },
    {
      "epoch": 8.079504594486616,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 20223
    },
    {
      "epoch": 8.079904115061925,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 20224
    },
    {
      "epoch": 8.080303635637236,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 20225
    },
    {
      "epoch": 8.080703156212545,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 20226
    },
    {
      "epoch": 8.081102676787854,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 20227
    },
    {
      "epoch": 8.081502197363164,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 20228
    },
    {
      "epoch": 8.081901717938473,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 20229
    },
    {
      "epoch": 8.082301238513784,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 20230
    },
    {
      "epoch": 8.082700759089093,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 20231
    },
    {
      "epoch": 8.083100279664404,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 20232
    },
    {
      "epoch": 8.083499800239712,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 20233
    },
    {
      "epoch": 8.083899320815021,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 20234
    },
    {
      "epoch": 8.084298841390332,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 20235
    },
    {
      "epoch": 8.084698361965641,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 20236
    },
    {
      "epoch": 8.085097882540952,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 20237
    },
    {
      "epoch": 8.08549740311626,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 20238
    },
    {
      "epoch": 8.08589692369157,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 20239
    },
    {
      "epoch": 8.08629644426688,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 20240
    },
    {
      "epoch": 8.086695964842189,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 20241
    },
    {
      "epoch": 8.0870954854175,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 20242
    },
    {
      "epoch": 8.087495005992809,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 20243
    },
    {
      "epoch": 8.087894526568117,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 20244
    },
    {
      "epoch": 8.088294047143428,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 20245
    },
    {
      "epoch": 8.088693567718737,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 20246
    },
    {
      "epoch": 8.089093088294048,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 20247
    },
    {
      "epoch": 8.089492608869357,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 20248
    },
    {
      "epoch": 8.089892129444667,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 20249
    },
    {
      "epoch": 8.090291650019976,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 20250
    },
    {
      "epoch": 8.090691170595285,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 20251
    },
    {
      "epoch": 8.091090691170596,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 20252
    },
    {
      "epoch": 8.091490211745905,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 20253
    },
    {
      "epoch": 8.091889732321215,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 20254
    },
    {
      "epoch": 8.092289252896524,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 20255
    },
    {
      "epoch": 8.092688773471833,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 20256
    },
    {
      "epoch": 8.093088294047144,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 20257
    },
    {
      "epoch": 8.093487814622453,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 20258
    },
    {
      "epoch": 8.093887335197763,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 20259
    },
    {
      "epoch": 8.094286855773072,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 20260
    },
    {
      "epoch": 8.094686376348381,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 20261
    },
    {
      "epoch": 8.095085896923692,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 20262
    },
    {
      "epoch": 8.095485417499,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 20263
    },
    {
      "epoch": 8.095884938074311,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 20264
    },
    {
      "epoch": 8.09628445864962,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 20265
    },
    {
      "epoch": 8.09668397922493,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 20266
    },
    {
      "epoch": 8.09708349980024,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 20267
    },
    {
      "epoch": 8.097483020375549,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 20268
    },
    {
      "epoch": 8.09788254095086,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 20269
    },
    {
      "epoch": 8.098282061526168,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 20270
    },
    {
      "epoch": 8.098681582101479,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 20271
    },
    {
      "epoch": 8.099081102676788,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 20272
    },
    {
      "epoch": 8.099480623252097,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 20273
    },
    {
      "epoch": 8.099880143827408,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 20274
    },
    {
      "epoch": 8.100279664402716,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 20275
    },
    {
      "epoch": 8.100679184978027,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 20276
    },
    {
      "epoch": 8.101078705553336,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 20277
    },
    {
      "epoch": 8.101478226128645,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 20278
    },
    {
      "epoch": 8.101877746703956,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 20279
    },
    {
      "epoch": 8.102277267279264,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 20280
    },
    {
      "epoch": 8.102676787854575,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 20281
    },
    {
      "epoch": 8.103076308429884,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 20282
    },
    {
      "epoch": 8.103475829005193,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 20283
    },
    {
      "epoch": 8.103875349580504,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 20284
    },
    {
      "epoch": 8.104274870155812,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 20285
    },
    {
      "epoch": 8.104674390731123,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 20286
    },
    {
      "epoch": 8.105073911306432,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 20287
    },
    {
      "epoch": 8.105473431881743,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 20288
    },
    {
      "epoch": 8.105872952457052,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 20289
    },
    {
      "epoch": 8.10627247303236,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 20290
    },
    {
      "epoch": 8.106671993607671,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 20291
    },
    {
      "epoch": 8.10707151418298,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 20292
    },
    {
      "epoch": 8.10747103475829,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 20293
    },
    {
      "epoch": 8.1078705553336,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 20294
    },
    {
      "epoch": 8.108270075908909,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 20295
    },
    {
      "epoch": 8.10866959648422,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 20296
    },
    {
      "epoch": 8.109069117059528,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 20297
    },
    {
      "epoch": 8.109468637634839,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 20298
    },
    {
      "epoch": 8.109868158210148,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 20299
    },
    {
      "epoch": 8.110267678785457,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 20300
    },
    {
      "epoch": 8.110667199360767,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 20301
    },
    {
      "epoch": 8.111066719936076,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 20302
    },
    {
      "epoch": 8.111466240511387,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 20303
    },
    {
      "epoch": 8.111865761086696,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 20304
    },
    {
      "epoch": 8.112265281662006,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 20305
    },
    {
      "epoch": 8.112664802237315,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 20306
    },
    {
      "epoch": 8.113064322812624,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 20307
    },
    {
      "epoch": 8.113463843387935,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 20308
    },
    {
      "epoch": 8.113863363963244,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 20309
    },
    {
      "epoch": 8.114262884538554,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 20310
    },
    {
      "epoch": 8.114662405113863,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 20311
    },
    {
      "epoch": 8.115061925689172,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 20312
    },
    {
      "epoch": 8.115461446264483,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 20313
    },
    {
      "epoch": 8.115860966839792,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 20314
    },
    {
      "epoch": 8.116260487415103,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 20315
    },
    {
      "epoch": 8.116660007990411,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 20316
    },
    {
      "epoch": 8.11705952856572,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 20317
    },
    {
      "epoch": 8.117459049141031,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 20318
    },
    {
      "epoch": 8.11785856971634,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 20319
    },
    {
      "epoch": 8.11825809029165,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 20320
    },
    {
      "epoch": 8.11865761086696,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 20321
    },
    {
      "epoch": 8.11905713144227,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 20322
    },
    {
      "epoch": 8.119456652017579,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 20323
    },
    {
      "epoch": 8.119856172592888,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 20324
    },
    {
      "epoch": 8.120255693168199,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 20325
    },
    {
      "epoch": 8.120655213743508,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 20326
    },
    {
      "epoch": 8.121054734318818,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 20327
    },
    {
      "epoch": 8.121454254894127,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 20328
    },
    {
      "epoch": 8.121853775469436,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 20329
    },
    {
      "epoch": 8.122253296044747,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 20330
    },
    {
      "epoch": 8.122652816620056,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 20331
    },
    {
      "epoch": 8.123052337195366,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 20332
    },
    {
      "epoch": 8.123451857770675,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 20333
    },
    {
      "epoch": 8.123851378345984,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 20334
    },
    {
      "epoch": 8.124250898921295,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 20335
    },
    {
      "epoch": 8.124650419496604,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 20336
    },
    {
      "epoch": 8.125049940071914,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 20337
    },
    {
      "epoch": 8.125449460647223,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 20338
    },
    {
      "epoch": 8.125848981222532,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 20339
    },
    {
      "epoch": 8.126248501797843,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 20340
    },
    {
      "epoch": 8.126648022373152,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 20341
    },
    {
      "epoch": 8.127047542948462,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 20342
    },
    {
      "epoch": 8.127447063523771,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 20343
    },
    {
      "epoch": 8.127846584099082,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 20344
    },
    {
      "epoch": 8.12824610467439,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 20345
    },
    {
      "epoch": 8.1286456252497,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 20346
    },
    {
      "epoch": 8.12904514582501,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 20347
    },
    {
      "epoch": 8.12944466640032,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 20348
    },
    {
      "epoch": 8.12984418697563,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 20349
    },
    {
      "epoch": 8.130243707550939,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 20350
    },
    {
      "epoch": 8.130643228126248,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 20351
    },
    {
      "epoch": 8.131042748701558,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 20352
    },
    {
      "epoch": 8.131442269276867,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 20353
    },
    {
      "epoch": 8.131841789852178,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 20354
    },
    {
      "epoch": 8.132241310427487,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 20355
    },
    {
      "epoch": 8.132640831002796,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 20356
    },
    {
      "epoch": 8.133040351578106,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 20357
    },
    {
      "epoch": 8.133439872153415,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 20358
    },
    {
      "epoch": 8.133839392728726,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 20359
    },
    {
      "epoch": 8.134238913304035,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 20360
    },
    {
      "epoch": 8.134638433879346,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 20361
    },
    {
      "epoch": 8.135037954454654,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 20362
    },
    {
      "epoch": 8.135437475029963,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 20363
    },
    {
      "epoch": 8.135836995605274,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 20364
    },
    {
      "epoch": 8.136236516180583,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 20365
    },
    {
      "epoch": 8.136636036755894,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 20366
    },
    {
      "epoch": 8.137035557331203,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 20367
    },
    {
      "epoch": 8.137435077906511,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 20368
    },
    {
      "epoch": 8.137834598481822,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 20369
    },
    {
      "epoch": 8.138234119057131,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 20370
    },
    {
      "epoch": 8.138633639632442,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 20371
    },
    {
      "epoch": 8.13903316020775,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 20372
    },
    {
      "epoch": 8.13943268078306,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 20373
    },
    {
      "epoch": 8.13983220135837,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 20374
    },
    {
      "epoch": 8.140231721933679,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 20375
    },
    {
      "epoch": 8.14063124250899,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 20376
    },
    {
      "epoch": 8.141030763084299,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 20377
    },
    {
      "epoch": 8.14143028365961,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 20378
    },
    {
      "epoch": 8.141829804234918,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 20379
    },
    {
      "epoch": 8.142229324810227,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 20380
    },
    {
      "epoch": 8.142628845385538,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 20381
    },
    {
      "epoch": 8.143028365960847,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 20382
    },
    {
      "epoch": 8.143427886536157,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 20383
    },
    {
      "epoch": 8.143827407111466,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 20384
    },
    {
      "epoch": 8.144226927686775,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 20385
    },
    {
      "epoch": 8.144626448262086,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 20386
    },
    {
      "epoch": 8.145025968837395,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 20387
    },
    {
      "epoch": 8.145425489412705,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 20388
    },
    {
      "epoch": 8.145825009988014,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 20389
    },
    {
      "epoch": 8.146224530563323,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 20390
    },
    {
      "epoch": 8.146624051138634,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 20391
    },
    {
      "epoch": 8.147023571713943,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 20392
    },
    {
      "epoch": 8.147423092289253,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 20393
    },
    {
      "epoch": 8.147822612864562,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 20394
    },
    {
      "epoch": 8.148222133439873,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 20395
    },
    {
      "epoch": 8.148621654015182,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 20396
    },
    {
      "epoch": 8.14902117459049,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 20397
    },
    {
      "epoch": 8.149420695165801,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 20398
    },
    {
      "epoch": 8.14982021574111,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 20399
    },
    {
      "epoch": 8.150219736316421,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 20400
    },
    {
      "epoch": 8.15061925689173,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 20401
    },
    {
      "epoch": 8.151018777467039,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 20402
    },
    {
      "epoch": 8.15141829804235,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 20403
    },
    {
      "epoch": 8.151817818617658,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 20404
    },
    {
      "epoch": 8.152217339192969,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 20405
    },
    {
      "epoch": 8.152616859768278,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 20406
    },
    {
      "epoch": 8.153016380343587,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 20407
    },
    {
      "epoch": 8.153415900918898,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 20408
    },
    {
      "epoch": 8.153815421494206,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 20409
    },
    {
      "epoch": 8.154214942069517,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 20410
    },
    {
      "epoch": 8.154614462644826,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 20411
    },
    {
      "epoch": 8.155013983220137,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 20412
    },
    {
      "epoch": 8.155413503795446,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 20413
    },
    {
      "epoch": 8.155813024370755,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 20414
    },
    {
      "epoch": 8.156212544946065,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 20415
    },
    {
      "epoch": 8.156612065521374,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 20416
    },
    {
      "epoch": 8.157011586096685,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 20417
    },
    {
      "epoch": 8.157411106671994,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 20418
    },
    {
      "epoch": 8.157810627247303,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 20419
    },
    {
      "epoch": 8.158210147822613,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 20420
    },
    {
      "epoch": 8.158609668397922,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 20421
    },
    {
      "epoch": 8.159009188973233,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 20422
    },
    {
      "epoch": 8.159408709548542,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 20423
    },
    {
      "epoch": 8.15980823012385,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 20424
    },
    {
      "epoch": 8.160207750699161,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 20425
    },
    {
      "epoch": 8.16060727127447,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 20426
    },
    {
      "epoch": 8.16100679184978,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 20427
    },
    {
      "epoch": 8.16140631242509,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 20428
    },
    {
      "epoch": 8.1618058330004,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 20429
    },
    {
      "epoch": 8.16220535357571,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 20430
    },
    {
      "epoch": 8.162604874151018,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 20431
    },
    {
      "epoch": 8.163004394726329,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 20432
    },
    {
      "epoch": 8.163403915301638,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 20433
    },
    {
      "epoch": 8.163803435876948,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 20434
    },
    {
      "epoch": 8.164202956452257,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 20435
    },
    {
      "epoch": 8.164602477027566,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 20436
    },
    {
      "epoch": 8.165001997602877,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 20437
    },
    {
      "epoch": 8.165401518178186,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 20438
    },
    {
      "epoch": 8.165801038753496,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 20439
    },
    {
      "epoch": 8.166200559328805,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 20440
    },
    {
      "epoch": 8.166600079904114,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 20441
    },
    {
      "epoch": 8.166999600479425,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 20442
    },
    {
      "epoch": 8.167399121054734,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 20443
    },
    {
      "epoch": 8.167798641630045,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 20444
    },
    {
      "epoch": 8.168198162205353,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 20445
    },
    {
      "epoch": 8.168597682780662,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 20446
    },
    {
      "epoch": 8.168997203355973,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 20447
    },
    {
      "epoch": 8.169396723931282,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 20448
    },
    {
      "epoch": 8.169796244506593,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 20449
    },
    {
      "epoch": 8.170195765081901,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 20450
    },
    {
      "epoch": 8.170595285657212,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 20451
    },
    {
      "epoch": 8.170994806232521,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 20452
    },
    {
      "epoch": 8.17139432680783,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 20453
    },
    {
      "epoch": 8.17179384738314,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 20454
    },
    {
      "epoch": 8.17219336795845,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 20455
    },
    {
      "epoch": 8.17259288853376,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 20456
    },
    {
      "epoch": 8.172992409109069,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 20457
    },
    {
      "epoch": 8.173391929684378,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 20458
    },
    {
      "epoch": 8.173791450259689,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 20459
    },
    {
      "epoch": 8.174190970834998,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 20460
    },
    {
      "epoch": 8.174590491410308,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 20461
    },
    {
      "epoch": 8.174990011985617,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 20462
    },
    {
      "epoch": 8.175389532560926,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 20463
    },
    {
      "epoch": 8.175789053136237,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 20464
    },
    {
      "epoch": 8.176188573711546,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 20465
    },
    {
      "epoch": 8.176588094286856,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 20466
    },
    {
      "epoch": 8.176987614862165,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 20467
    },
    {
      "epoch": 8.177387135437476,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 20468
    },
    {
      "epoch": 8.177786656012785,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 20469
    },
    {
      "epoch": 8.178186176588094,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 20470
    },
    {
      "epoch": 8.178585697163404,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 20471
    },
    {
      "epoch": 8.178985217738713,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 20472
    },
    {
      "epoch": 8.179384738314024,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 20473
    },
    {
      "epoch": 8.179784258889333,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 20474
    },
    {
      "epoch": 8.180183779464642,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 20475
    },
    {
      "epoch": 8.180583300039952,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 20476
    },
    {
      "epoch": 8.180982820615261,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 20477
    },
    {
      "epoch": 8.181382341190572,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0578,
      "step": 20478
    },
    {
      "epoch": 8.18178186176588,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 20479
    },
    {
      "epoch": 8.18218138234119,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 20480
    },
    {
      "epoch": 8.1825809029165,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 20481
    },
    {
      "epoch": 8.18298042349181,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 20482
    },
    {
      "epoch": 8.18337994406712,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 20483
    },
    {
      "epoch": 8.183779464642429,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 20484
    },
    {
      "epoch": 8.18417898521774,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 20485
    },
    {
      "epoch": 8.184578505793048,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 20486
    },
    {
      "epoch": 8.184978026368357,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 20487
    },
    {
      "epoch": 8.185377546943668,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 20488
    },
    {
      "epoch": 8.185777067518977,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 20489
    },
    {
      "epoch": 8.186176588094288,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 20490
    },
    {
      "epoch": 8.186576108669597,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 20491
    },
    {
      "epoch": 8.186975629244905,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 20492
    },
    {
      "epoch": 8.187375149820216,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 20493
    },
    {
      "epoch": 8.187774670395525,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 20494
    },
    {
      "epoch": 8.188174190970836,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 20495
    },
    {
      "epoch": 8.188573711546145,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 20496
    },
    {
      "epoch": 8.188973232121453,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 20497
    },
    {
      "epoch": 8.189372752696764,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 20498
    },
    {
      "epoch": 8.189772273272073,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 20499
    },
    {
      "epoch": 8.190171793847384,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 20500
    },
    {
      "epoch": 8.190571314422693,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 20501
    },
    {
      "epoch": 8.190970834998003,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 20502
    },
    {
      "epoch": 8.191370355573312,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 20503
    },
    {
      "epoch": 8.191769876148621,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 20504
    },
    {
      "epoch": 8.192169396723932,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 20505
    },
    {
      "epoch": 8.19256891729924,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 20506
    },
    {
      "epoch": 8.192968437874551,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 20507
    },
    {
      "epoch": 8.19336795844986,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 20508
    },
    {
      "epoch": 8.193767479025169,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 20509
    },
    {
      "epoch": 8.19416699960048,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 20510
    },
    {
      "epoch": 8.194566520175789,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 20511
    },
    {
      "epoch": 8.1949660407511,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 20512
    },
    {
      "epoch": 8.195365561326408,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 20513
    },
    {
      "epoch": 8.195765081901717,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 20514
    },
    {
      "epoch": 8.196164602477028,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 20515
    },
    {
      "epoch": 8.196564123052337,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 20516
    },
    {
      "epoch": 8.196963643627647,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 20517
    },
    {
      "epoch": 8.197363164202956,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 20518
    },
    {
      "epoch": 8.197762684778265,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 20519
    },
    {
      "epoch": 8.198162205353576,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 20520
    },
    {
      "epoch": 8.198561725928885,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 20521
    },
    {
      "epoch": 8.198961246504195,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 20522
    },
    {
      "epoch": 8.199360767079504,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 20523
    },
    {
      "epoch": 8.199760287654815,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 20524
    },
    {
      "epoch": 8.200159808230124,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 20525
    },
    {
      "epoch": 8.200559328805433,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 20526
    },
    {
      "epoch": 8.200958849380743,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 20527
    },
    {
      "epoch": 8.201358369956052,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 20528
    },
    {
      "epoch": 8.201757890531363,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 20529
    },
    {
      "epoch": 8.202157411106672,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 20530
    },
    {
      "epoch": 8.20255693168198,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 20531
    },
    {
      "epoch": 8.202956452257292,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 20532
    },
    {
      "epoch": 8.2033559728326,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 20533
    },
    {
      "epoch": 8.203755493407911,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 20534
    },
    {
      "epoch": 8.20415501398322,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 20535
    },
    {
      "epoch": 8.204554534558529,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 20536
    },
    {
      "epoch": 8.20495405513384,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 20537
    },
    {
      "epoch": 8.205353575709148,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 20538
    },
    {
      "epoch": 8.20575309628446,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 20539
    },
    {
      "epoch": 8.206152616859768,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 20540
    },
    {
      "epoch": 8.206552137435079,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 20541
    },
    {
      "epoch": 8.206951658010388,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 20542
    },
    {
      "epoch": 8.207351178585697,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 20543
    },
    {
      "epoch": 8.207750699161007,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 20544
    },
    {
      "epoch": 8.208150219736316,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 20545
    },
    {
      "epoch": 8.208549740311627,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 20546
    },
    {
      "epoch": 8.208949260886936,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 20547
    },
    {
      "epoch": 8.209348781462245,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 20548
    },
    {
      "epoch": 8.209748302037555,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 20549
    },
    {
      "epoch": 8.210147822612864,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 20550
    },
    {
      "epoch": 8.210547343188175,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 20551
    },
    {
      "epoch": 8.210946863763484,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 20552
    },
    {
      "epoch": 8.211346384338793,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 20553
    },
    {
      "epoch": 8.211745904914103,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 20554
    },
    {
      "epoch": 8.212145425489412,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 20555
    },
    {
      "epoch": 8.212544946064723,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 20556
    },
    {
      "epoch": 8.212944466640032,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 20557
    },
    {
      "epoch": 8.213343987215342,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 20558
    },
    {
      "epoch": 8.213743507790651,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 20559
    },
    {
      "epoch": 8.21414302836596,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 20560
    },
    {
      "epoch": 8.214542548941271,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 20561
    },
    {
      "epoch": 8.21494206951658,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 20562
    },
    {
      "epoch": 8.21534159009189,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 20563
    },
    {
      "epoch": 8.2157411106672,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 20564
    },
    {
      "epoch": 8.216140631242508,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 20565
    },
    {
      "epoch": 8.216540151817819,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 20566
    },
    {
      "epoch": 8.216939672393128,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 20567
    },
    {
      "epoch": 8.217339192968439,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 20568
    },
    {
      "epoch": 8.217738713543747,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 20569
    },
    {
      "epoch": 8.218138234119056,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 20570
    },
    {
      "epoch": 8.218537754694367,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 20571
    },
    {
      "epoch": 8.218937275269676,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 20572
    },
    {
      "epoch": 8.219336795844987,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 20573
    },
    {
      "epoch": 8.219736316420295,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 20574
    },
    {
      "epoch": 8.220135836995606,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0546,
      "step": 20575
    },
    {
      "epoch": 8.220535357570915,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 20576
    },
    {
      "epoch": 8.220934878146224,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 20577
    },
    {
      "epoch": 8.221334398721535,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 20578
    },
    {
      "epoch": 8.221733919296843,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 20579
    },
    {
      "epoch": 8.222133439872154,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 20580
    },
    {
      "epoch": 8.222532960447463,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 20581
    },
    {
      "epoch": 8.222932481022772,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 20582
    },
    {
      "epoch": 8.223332001598083,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 20583
    },
    {
      "epoch": 8.223731522173392,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 20584
    },
    {
      "epoch": 8.224131042748702,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 20585
    },
    {
      "epoch": 8.224530563324011,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 20586
    },
    {
      "epoch": 8.22493008389932,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 20587
    },
    {
      "epoch": 8.22532960447463,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 20588
    },
    {
      "epoch": 8.22572912504994,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 20589
    },
    {
      "epoch": 8.22612864562525,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 20590
    },
    {
      "epoch": 8.22652816620056,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 20591
    },
    {
      "epoch": 8.22692768677587,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 20592
    },
    {
      "epoch": 8.227327207351179,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 20593
    },
    {
      "epoch": 8.227726727926488,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 20594
    },
    {
      "epoch": 8.228126248501798,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 20595
    },
    {
      "epoch": 8.228525769077107,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 20596
    },
    {
      "epoch": 8.228925289652418,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 20597
    },
    {
      "epoch": 8.229324810227727,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 20598
    },
    {
      "epoch": 8.229724330803036,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 20599
    },
    {
      "epoch": 8.230123851378346,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 20600
    },
    {
      "epoch": 8.230523371953655,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 20601
    },
    {
      "epoch": 8.230922892528966,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 20602
    },
    {
      "epoch": 8.231322413104275,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 20603
    },
    {
      "epoch": 8.231721933679584,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 20604
    },
    {
      "epoch": 8.232121454254894,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 20605
    },
    {
      "epoch": 8.232520974830203,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 20606
    },
    {
      "epoch": 8.232920495405514,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 20607
    },
    {
      "epoch": 8.233320015980823,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 20608
    },
    {
      "epoch": 8.233719536556134,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 20609
    },
    {
      "epoch": 8.234119057131442,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 20610
    },
    {
      "epoch": 8.234518577706751,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 20611
    },
    {
      "epoch": 8.234918098282062,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 20612
    },
    {
      "epoch": 8.235317618857371,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 20613
    },
    {
      "epoch": 8.235717139432682,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.055,
      "step": 20614
    },
    {
      "epoch": 8.23611666000799,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 20615
    },
    {
      "epoch": 8.2365161805833,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 20616
    },
    {
      "epoch": 8.23691570115861,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 20617
    },
    {
      "epoch": 8.237315221733919,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 20618
    },
    {
      "epoch": 8.23771474230923,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 20619
    },
    {
      "epoch": 8.238114262884539,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 20620
    },
    {
      "epoch": 8.238513783459847,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 20621
    },
    {
      "epoch": 8.238913304035158,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 20622
    },
    {
      "epoch": 8.239312824610467,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 20623
    },
    {
      "epoch": 8.239712345185778,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 20624
    },
    {
      "epoch": 8.240111865761087,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 20625
    },
    {
      "epoch": 8.240511386336395,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 20626
    },
    {
      "epoch": 8.240910906911706,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 20627
    },
    {
      "epoch": 8.241310427487015,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 20628
    },
    {
      "epoch": 8.241709948062326,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 20629
    },
    {
      "epoch": 8.242109468637635,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 20630
    },
    {
      "epoch": 8.242508989212945,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 20631
    },
    {
      "epoch": 8.242908509788254,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 20632
    },
    {
      "epoch": 8.243308030363563,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 20633
    },
    {
      "epoch": 8.243707550938874,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 20634
    },
    {
      "epoch": 8.244107071514183,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 20635
    },
    {
      "epoch": 8.244506592089493,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 20636
    },
    {
      "epoch": 8.244906112664802,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 20637
    },
    {
      "epoch": 8.245305633240111,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 20638
    },
    {
      "epoch": 8.245705153815422,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 20639
    },
    {
      "epoch": 8.24610467439073,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 20640
    },
    {
      "epoch": 8.246504194966041,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 20641
    },
    {
      "epoch": 8.24690371554135,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 20642
    },
    {
      "epoch": 8.24730323611666,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 20643
    },
    {
      "epoch": 8.24770275669197,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 20644
    },
    {
      "epoch": 8.248102277267279,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 20645
    },
    {
      "epoch": 8.24850179784259,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 20646
    },
    {
      "epoch": 8.248901318417898,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 20647
    },
    {
      "epoch": 8.249300838993209,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 20648
    },
    {
      "epoch": 8.249700359568518,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 20649
    },
    {
      "epoch": 8.250099880143827,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 20650
    },
    {
      "epoch": 8.250499400719137,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 20651
    },
    {
      "epoch": 8.250898921294446,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 20652
    },
    {
      "epoch": 8.251298441869757,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 20653
    },
    {
      "epoch": 8.251697962445066,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 20654
    },
    {
      "epoch": 8.252097483020375,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 20655
    },
    {
      "epoch": 8.252497003595685,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 20656
    },
    {
      "epoch": 8.252896524170994,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 20657
    },
    {
      "epoch": 8.253296044746305,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 20658
    },
    {
      "epoch": 8.253695565321614,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 20659
    },
    {
      "epoch": 8.254095085896923,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 20660
    },
    {
      "epoch": 8.254494606472234,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 20661
    },
    {
      "epoch": 8.254894127047542,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 20662
    },
    {
      "epoch": 8.255293647622853,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 20663
    },
    {
      "epoch": 8.255693168198162,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 20664
    },
    {
      "epoch": 8.256092688773473,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 20665
    },
    {
      "epoch": 8.256492209348782,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 20666
    },
    {
      "epoch": 8.25689172992409,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 20667
    },
    {
      "epoch": 8.257291250499401,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 20668
    },
    {
      "epoch": 8.25769077107471,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 20669
    },
    {
      "epoch": 8.25809029165002,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 20670
    },
    {
      "epoch": 8.25848981222533,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 20671
    },
    {
      "epoch": 8.258889332800639,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 20672
    },
    {
      "epoch": 8.25928885337595,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 20673
    },
    {
      "epoch": 8.259688373951258,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 20674
    },
    {
      "epoch": 8.260087894526569,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 20675
    },
    {
      "epoch": 8.260487415101878,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 20676
    },
    {
      "epoch": 8.260886935677187,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 20677
    },
    {
      "epoch": 8.261286456252497,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 20678
    },
    {
      "epoch": 8.261685976827806,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 20679
    },
    {
      "epoch": 8.262085497403117,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 20680
    },
    {
      "epoch": 8.262485017978426,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 20681
    },
    {
      "epoch": 8.262884538553735,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 20682
    },
    {
      "epoch": 8.263284059129045,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 20683
    },
    {
      "epoch": 8.263683579704354,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 20684
    },
    {
      "epoch": 8.264083100279665,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 20685
    },
    {
      "epoch": 8.264482620854974,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 20686
    },
    {
      "epoch": 8.264882141430284,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 20687
    },
    {
      "epoch": 8.265281662005593,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 20688
    },
    {
      "epoch": 8.265681182580902,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0541,
      "step": 20689
    },
    {
      "epoch": 8.266080703156213,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 20690
    },
    {
      "epoch": 8.266480223731522,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 20691
    },
    {
      "epoch": 8.266879744306832,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 20692
    },
    {
      "epoch": 8.267279264882141,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 20693
    },
    {
      "epoch": 8.26767878545745,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 20694
    },
    {
      "epoch": 8.268078306032761,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 20695
    },
    {
      "epoch": 8.26847782660807,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 20696
    },
    {
      "epoch": 8.26887734718338,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 20697
    },
    {
      "epoch": 8.26927686775869,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 20698
    },
    {
      "epoch": 8.269676388333998,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 20699
    },
    {
      "epoch": 8.270075908909309,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 20700
    },
    {
      "epoch": 8.270475429484618,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 20701
    },
    {
      "epoch": 8.270874950059929,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 20702
    },
    {
      "epoch": 8.271274470635237,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 20703
    },
    {
      "epoch": 8.271673991210548,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 20704
    },
    {
      "epoch": 8.272073511785857,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 20705
    },
    {
      "epoch": 8.272473032361166,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 20706
    },
    {
      "epoch": 8.272872552936477,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 20707
    },
    {
      "epoch": 8.273272073511786,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 20708
    },
    {
      "epoch": 8.273671594087096,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 20709
    },
    {
      "epoch": 8.274071114662405,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 20710
    },
    {
      "epoch": 8.274470635237714,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 20711
    },
    {
      "epoch": 8.274870155813025,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 20712
    },
    {
      "epoch": 8.275269676388334,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 20713
    },
    {
      "epoch": 8.275669196963644,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 20714
    },
    {
      "epoch": 8.276068717538953,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 20715
    },
    {
      "epoch": 8.276468238114262,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 20716
    },
    {
      "epoch": 8.276867758689573,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 20717
    },
    {
      "epoch": 8.277267279264882,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 20718
    },
    {
      "epoch": 8.277666799840192,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 20719
    },
    {
      "epoch": 8.278066320415501,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 20720
    },
    {
      "epoch": 8.278465840990812,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 20721
    },
    {
      "epoch": 8.27886536156612,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 20722
    },
    {
      "epoch": 8.27926488214143,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 20723
    },
    {
      "epoch": 8.27966440271674,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 20724
    },
    {
      "epoch": 8.28006392329205,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 20725
    },
    {
      "epoch": 8.28046344386736,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 20726
    },
    {
      "epoch": 8.280862964442669,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 20727
    },
    {
      "epoch": 8.281262485017978,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 20728
    },
    {
      "epoch": 8.281662005593288,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 20729
    },
    {
      "epoch": 8.282061526168597,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 20730
    },
    {
      "epoch": 8.282461046743908,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 20731
    },
    {
      "epoch": 8.282860567319217,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 20732
    },
    {
      "epoch": 8.283260087894526,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 20733
    },
    {
      "epoch": 8.283659608469836,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 20734
    },
    {
      "epoch": 8.284059129045145,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 20735
    },
    {
      "epoch": 8.284458649620456,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 20736
    },
    {
      "epoch": 8.284858170195765,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 20737
    },
    {
      "epoch": 8.285257690771076,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 20738
    },
    {
      "epoch": 8.285657211346384,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 20739
    },
    {
      "epoch": 8.286056731921693,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 20740
    },
    {
      "epoch": 8.286456252497004,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 20741
    },
    {
      "epoch": 8.286855773072313,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 20742
    },
    {
      "epoch": 8.287255293647624,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 20743
    },
    {
      "epoch": 8.287654814222932,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 20744
    },
    {
      "epoch": 8.288054334798241,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 20745
    },
    {
      "epoch": 8.288453855373552,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 20746
    },
    {
      "epoch": 8.288853375948861,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 20747
    },
    {
      "epoch": 8.289252896524172,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 20748
    },
    {
      "epoch": 8.28965241709948,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 20749
    },
    {
      "epoch": 8.29005193767479,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 20750
    },
    {
      "epoch": 8.2904514582501,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 20751
    },
    {
      "epoch": 8.290850978825409,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 20752
    },
    {
      "epoch": 8.29125049940072,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 20753
    },
    {
      "epoch": 8.291650019976029,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 20754
    },
    {
      "epoch": 8.29204954055134,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 20755
    },
    {
      "epoch": 8.292449061126648,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 20756
    },
    {
      "epoch": 8.292848581701957,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 20757
    },
    {
      "epoch": 8.293248102277268,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 20758
    },
    {
      "epoch": 8.293647622852577,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 20759
    },
    {
      "epoch": 8.294047143427887,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 20760
    },
    {
      "epoch": 8.294446664003196,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 20761
    },
    {
      "epoch": 8.294846184578505,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 20762
    },
    {
      "epoch": 8.295245705153816,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 20763
    },
    {
      "epoch": 8.295645225729125,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 20764
    },
    {
      "epoch": 8.296044746304435,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 20765
    },
    {
      "epoch": 8.296444266879744,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 20766
    },
    {
      "epoch": 8.296843787455053,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 20767
    },
    {
      "epoch": 8.297243308030364,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 20768
    },
    {
      "epoch": 8.297642828605673,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 20769
    },
    {
      "epoch": 8.298042349180983,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 20770
    },
    {
      "epoch": 8.298441869756292,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 20771
    },
    {
      "epoch": 8.298841390331603,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 20772
    },
    {
      "epoch": 8.299240910906912,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 20773
    },
    {
      "epoch": 8.29964043148222,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 20774
    },
    {
      "epoch": 8.300039952057531,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 20775
    },
    {
      "epoch": 8.30043947263284,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 20776
    },
    {
      "epoch": 8.300838993208151,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 20777
    },
    {
      "epoch": 8.30123851378346,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 20778
    },
    {
      "epoch": 8.301638034358769,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 20779
    },
    {
      "epoch": 8.30203755493408,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 20780
    },
    {
      "epoch": 8.302437075509388,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 20781
    },
    {
      "epoch": 8.302836596084699,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 20782
    },
    {
      "epoch": 8.303236116660008,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 20783
    },
    {
      "epoch": 8.303635637235317,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 20784
    },
    {
      "epoch": 8.304035157810628,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 20785
    },
    {
      "epoch": 8.304434678385936,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 20786
    },
    {
      "epoch": 8.304834198961247,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 20787
    },
    {
      "epoch": 8.305233719536556,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 20788
    },
    {
      "epoch": 8.305633240111867,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 20789
    },
    {
      "epoch": 8.306032760687176,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 20790
    },
    {
      "epoch": 8.306432281262484,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 20791
    },
    {
      "epoch": 8.306831801837795,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 20792
    },
    {
      "epoch": 8.307231322413104,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 20793
    },
    {
      "epoch": 8.307630842988415,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 20794
    },
    {
      "epoch": 8.308030363563724,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 20795
    },
    {
      "epoch": 8.308429884139032,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 20796
    },
    {
      "epoch": 8.308829404714343,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 20797
    },
    {
      "epoch": 8.309228925289652,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 20798
    },
    {
      "epoch": 8.309628445864963,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 20799
    },
    {
      "epoch": 8.310027966440272,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 20800
    },
    {
      "epoch": 8.31042748701558,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 20801
    },
    {
      "epoch": 8.310827007590891,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 20802
    },
    {
      "epoch": 8.3112265281662,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 20803
    },
    {
      "epoch": 8.31162604874151,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 20804
    },
    {
      "epoch": 8.31202556931682,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 20805
    },
    {
      "epoch": 8.31242508989213,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 20806
    },
    {
      "epoch": 8.31282461046744,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 20807
    },
    {
      "epoch": 8.313224131042748,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 20808
    },
    {
      "epoch": 8.313623651618059,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 20809
    },
    {
      "epoch": 8.314023172193368,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 20810
    },
    {
      "epoch": 8.314422692768678,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 20811
    },
    {
      "epoch": 8.314822213343987,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 20812
    },
    {
      "epoch": 8.315221733919296,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 20813
    },
    {
      "epoch": 8.315621254494607,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 20814
    },
    {
      "epoch": 8.316020775069916,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 20815
    },
    {
      "epoch": 8.316420295645226,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 20816
    },
    {
      "epoch": 8.316819816220535,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 20817
    },
    {
      "epoch": 8.317219336795844,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0589,
      "step": 20818
    },
    {
      "epoch": 8.317618857371155,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 20819
    },
    {
      "epoch": 8.318018377946464,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 20820
    },
    {
      "epoch": 8.318417898521774,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 20821
    },
    {
      "epoch": 8.318817419097083,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 20822
    },
    {
      "epoch": 8.319216939672392,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 20823
    },
    {
      "epoch": 8.319616460247703,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 20824
    },
    {
      "epoch": 8.320015980823012,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 20825
    },
    {
      "epoch": 8.320415501398323,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 20826
    },
    {
      "epoch": 8.320815021973631,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 20827
    },
    {
      "epoch": 8.321214542548942,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 20828
    },
    {
      "epoch": 8.321614063124251,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 20829
    },
    {
      "epoch": 8.32201358369956,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 20830
    },
    {
      "epoch": 8.32241310427487,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 20831
    },
    {
      "epoch": 8.32281262485018,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 20832
    },
    {
      "epoch": 8.32321214542549,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 20833
    },
    {
      "epoch": 8.323611666000799,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 20834
    },
    {
      "epoch": 8.324011186576108,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 20835
    },
    {
      "epoch": 8.324410707151419,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 20836
    },
    {
      "epoch": 8.324810227726728,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 20837
    },
    {
      "epoch": 8.325209748302038,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 20838
    },
    {
      "epoch": 8.325609268877347,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 20839
    },
    {
      "epoch": 8.326008789452656,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 20840
    },
    {
      "epoch": 8.326408310027967,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 20841
    },
    {
      "epoch": 8.326807830603276,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 20842
    },
    {
      "epoch": 8.327207351178586,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 20843
    },
    {
      "epoch": 8.327606871753895,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 20844
    },
    {
      "epoch": 8.328006392329206,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 20845
    },
    {
      "epoch": 8.328405912904515,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 20846
    },
    {
      "epoch": 8.328805433479824,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 20847
    },
    {
      "epoch": 8.329204954055134,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 20848
    },
    {
      "epoch": 8.329604474630443,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 20849
    },
    {
      "epoch": 8.330003995205754,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 20850
    },
    {
      "epoch": 8.330403515781063,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 20851
    },
    {
      "epoch": 8.330803036356372,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 20852
    },
    {
      "epoch": 8.331202556931682,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 20853
    },
    {
      "epoch": 8.331602077506991,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 20854
    },
    {
      "epoch": 8.332001598082302,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 20855
    },
    {
      "epoch": 8.33240111865761,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 20856
    },
    {
      "epoch": 8.33280063923292,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 20857
    },
    {
      "epoch": 8.33320015980823,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 20858
    },
    {
      "epoch": 8.33359968038354,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 20859
    },
    {
      "epoch": 8.33399920095885,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 20860
    },
    {
      "epoch": 8.334398721534159,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 20861
    },
    {
      "epoch": 8.334798242109468,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 20862
    },
    {
      "epoch": 8.335197762684778,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 20863
    },
    {
      "epoch": 8.335597283260087,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 20864
    },
    {
      "epoch": 8.335996803835398,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0583,
      "step": 20865
    },
    {
      "epoch": 8.336396324410707,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 20866
    },
    {
      "epoch": 8.336795844986018,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 20867
    },
    {
      "epoch": 8.337195365561326,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 20868
    },
    {
      "epoch": 8.337594886136635,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 20869
    },
    {
      "epoch": 8.337994406711946,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 20870
    },
    {
      "epoch": 8.338393927287255,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 20871
    },
    {
      "epoch": 8.338793447862566,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 20872
    },
    {
      "epoch": 8.339192968437874,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 20873
    },
    {
      "epoch": 8.339592489013183,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 20874
    },
    {
      "epoch": 8.339992009588494,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 20875
    },
    {
      "epoch": 8.340391530163803,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 20876
    },
    {
      "epoch": 8.340791050739114,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 20877
    },
    {
      "epoch": 8.341190571314423,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 20878
    },
    {
      "epoch": 8.341590091889731,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 20879
    },
    {
      "epoch": 8.341989612465042,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 20880
    },
    {
      "epoch": 8.342389133040351,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 20881
    },
    {
      "epoch": 8.342788653615662,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 20882
    },
    {
      "epoch": 8.34318817419097,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 20883
    },
    {
      "epoch": 8.343587694766281,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 20884
    },
    {
      "epoch": 8.34398721534159,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 20885
    },
    {
      "epoch": 8.344386735916899,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 20886
    },
    {
      "epoch": 8.34478625649221,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 20887
    },
    {
      "epoch": 8.345185777067519,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 20888
    },
    {
      "epoch": 8.34558529764283,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 20889
    },
    {
      "epoch": 8.345984818218138,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 20890
    },
    {
      "epoch": 8.346384338793447,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 20891
    },
    {
      "epoch": 8.346783859368758,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 20892
    },
    {
      "epoch": 8.347183379944067,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 20893
    },
    {
      "epoch": 8.347582900519377,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 20894
    },
    {
      "epoch": 8.347982421094686,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 20895
    },
    {
      "epoch": 8.348381941669995,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 20896
    },
    {
      "epoch": 8.348781462245306,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 20897
    },
    {
      "epoch": 8.349180982820615,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 20898
    },
    {
      "epoch": 8.349580503395925,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 20899
    },
    {
      "epoch": 8.349980023971234,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 20900
    },
    {
      "epoch": 8.350379544546545,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 20901
    },
    {
      "epoch": 8.350779065121854,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 20902
    },
    {
      "epoch": 8.351178585697163,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 20903
    },
    {
      "epoch": 8.351578106272473,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 20904
    },
    {
      "epoch": 8.351977626847782,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 20905
    },
    {
      "epoch": 8.352377147423093,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 20906
    },
    {
      "epoch": 8.352776667998402,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 20907
    },
    {
      "epoch": 8.35317618857371,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 20908
    },
    {
      "epoch": 8.353575709149021,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 20909
    },
    {
      "epoch": 8.35397522972433,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 20910
    },
    {
      "epoch": 8.354374750299641,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 20911
    },
    {
      "epoch": 8.35477427087495,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 20912
    },
    {
      "epoch": 8.355173791450259,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 20913
    },
    {
      "epoch": 8.35557331202557,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 20914
    },
    {
      "epoch": 8.355972832600878,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 20915
    },
    {
      "epoch": 8.356372353176189,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 20916
    },
    {
      "epoch": 8.356771873751498,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 20917
    },
    {
      "epoch": 8.357171394326809,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 20918
    },
    {
      "epoch": 8.357570914902118,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 20919
    },
    {
      "epoch": 8.357970435477426,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 20920
    },
    {
      "epoch": 8.358369956052737,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 20921
    },
    {
      "epoch": 8.358769476628046,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 20922
    },
    {
      "epoch": 8.359168997203357,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 20923
    },
    {
      "epoch": 8.359568517778666,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 20924
    },
    {
      "epoch": 8.359968038353975,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 20925
    },
    {
      "epoch": 8.360367558929285,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 20926
    },
    {
      "epoch": 8.360767079504594,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 20927
    },
    {
      "epoch": 8.361166600079905,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 20928
    },
    {
      "epoch": 8.361566120655214,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 20929
    },
    {
      "epoch": 8.361965641230523,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 20930
    },
    {
      "epoch": 8.362365161805833,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 20931
    },
    {
      "epoch": 8.362764682381142,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 20932
    },
    {
      "epoch": 8.363164202956453,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 20933
    },
    {
      "epoch": 8.363563723531762,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 20934
    },
    {
      "epoch": 8.363963244107072,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 20935
    },
    {
      "epoch": 8.364362764682381,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 20936
    },
    {
      "epoch": 8.36476228525769,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 20937
    },
    {
      "epoch": 8.365161805833,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 20938
    },
    {
      "epoch": 8.36556132640831,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 20939
    },
    {
      "epoch": 8.36596084698362,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 20940
    },
    {
      "epoch": 8.36636036755893,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 20941
    },
    {
      "epoch": 8.366759888134238,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 20942
    },
    {
      "epoch": 8.367159408709549,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 20943
    },
    {
      "epoch": 8.367558929284858,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 20944
    },
    {
      "epoch": 8.367958449860168,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 20945
    },
    {
      "epoch": 8.368357970435477,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 20946
    },
    {
      "epoch": 8.368757491010786,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 20947
    },
    {
      "epoch": 8.369157011586097,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 20948
    },
    {
      "epoch": 8.369556532161406,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 20949
    },
    {
      "epoch": 8.369956052736717,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 20950
    },
    {
      "epoch": 8.370355573312025,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 20951
    },
    {
      "epoch": 8.370755093887336,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 20952
    },
    {
      "epoch": 8.371154614462645,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 20953
    },
    {
      "epoch": 8.371554135037954,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 20954
    },
    {
      "epoch": 8.371953655613265,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 20955
    },
    {
      "epoch": 8.372353176188573,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 20956
    },
    {
      "epoch": 8.372752696763884,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 20957
    },
    {
      "epoch": 8.373152217339193,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 20958
    },
    {
      "epoch": 8.373551737914502,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 20959
    },
    {
      "epoch": 8.373951258489813,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 20960
    },
    {
      "epoch": 8.374350779065121,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 20961
    },
    {
      "epoch": 8.374750299640432,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 20962
    },
    {
      "epoch": 8.375149820215741,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 20963
    },
    {
      "epoch": 8.37554934079105,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 20964
    },
    {
      "epoch": 8.37594886136636,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 20965
    },
    {
      "epoch": 8.37634838194167,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 20966
    },
    {
      "epoch": 8.37674790251698,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 20967
    },
    {
      "epoch": 8.377147423092289,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 20968
    },
    {
      "epoch": 8.3775469436676,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 20969
    },
    {
      "epoch": 8.377946464242909,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 20970
    },
    {
      "epoch": 8.378345984818218,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 20971
    },
    {
      "epoch": 8.378745505393528,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 20972
    },
    {
      "epoch": 8.379145025968837,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 20973
    },
    {
      "epoch": 8.379544546544148,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 20974
    },
    {
      "epoch": 8.379944067119457,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 20975
    },
    {
      "epoch": 8.380343587694766,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 20976
    },
    {
      "epoch": 8.380743108270076,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 20977
    },
    {
      "epoch": 8.381142628845385,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 20978
    },
    {
      "epoch": 8.381542149420696,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 20979
    },
    {
      "epoch": 8.381941669996005,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 20980
    },
    {
      "epoch": 8.382341190571314,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 20981
    },
    {
      "epoch": 8.382740711146624,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 20982
    },
    {
      "epoch": 8.383140231721933,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 20983
    },
    {
      "epoch": 8.383539752297244,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 20984
    },
    {
      "epoch": 8.383939272872553,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 20985
    },
    {
      "epoch": 8.384338793447863,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 20986
    },
    {
      "epoch": 8.384738314023172,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 20987
    },
    {
      "epoch": 8.385137834598481,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 20988
    },
    {
      "epoch": 8.385537355173792,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 20989
    },
    {
      "epoch": 8.3859368757491,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 20990
    },
    {
      "epoch": 8.386336396324412,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 20991
    },
    {
      "epoch": 8.38673591689972,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 20992
    },
    {
      "epoch": 8.38713543747503,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 20993
    },
    {
      "epoch": 8.38753495805034,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 20994
    },
    {
      "epoch": 8.387934478625649,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 20995
    },
    {
      "epoch": 8.38833399920096,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 20996
    },
    {
      "epoch": 8.388733519776268,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 20997
    },
    {
      "epoch": 8.389133040351577,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 20998
    },
    {
      "epoch": 8.389532560926888,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 20999
    },
    {
      "epoch": 8.389932081502197,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 21000
    },
    {
      "epoch": 8.390331602077508,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 21001
    },
    {
      "epoch": 8.390731122652817,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 21002
    },
    {
      "epoch": 8.391130643228125,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 21003
    },
    {
      "epoch": 8.391530163803436,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 21004
    },
    {
      "epoch": 8.391929684378745,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 21005
    },
    {
      "epoch": 8.392329204954056,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 21006
    },
    {
      "epoch": 8.392728725529365,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 21007
    },
    {
      "epoch": 8.393128246104675,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 21008
    },
    {
      "epoch": 8.393527766679984,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 21009
    },
    {
      "epoch": 8.393927287255293,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 21010
    },
    {
      "epoch": 8.394326807830604,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 21011
    },
    {
      "epoch": 8.394726328405913,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 21012
    },
    {
      "epoch": 8.395125848981223,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 21013
    },
    {
      "epoch": 8.395525369556532,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 21014
    },
    {
      "epoch": 8.395924890131841,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 21015
    },
    {
      "epoch": 8.396324410707152,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 21016
    },
    {
      "epoch": 8.39672393128246,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 21017
    },
    {
      "epoch": 8.397123451857771,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 21018
    },
    {
      "epoch": 8.39752297243308,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 21019
    },
    {
      "epoch": 8.397922493008389,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0085,
      "step": 21020
    },
    {
      "epoch": 8.3983220135837,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 21021
    },
    {
      "epoch": 8.398721534159009,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 21022
    },
    {
      "epoch": 8.39912105473432,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 21023
    },
    {
      "epoch": 8.399520575309628,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 21024
    },
    {
      "epoch": 8.399920095884939,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 21025
    },
    {
      "epoch": 8.400319616460248,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 21026
    },
    {
      "epoch": 8.400719137035557,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 21027
    },
    {
      "epoch": 8.401118657610867,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 21028
    },
    {
      "epoch": 8.401518178186176,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 21029
    },
    {
      "epoch": 8.401917698761487,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 21030
    },
    {
      "epoch": 8.402317219336796,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 21031
    },
    {
      "epoch": 8.402716739912105,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 21032
    },
    {
      "epoch": 8.403116260487415,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 21033
    },
    {
      "epoch": 8.403515781062724,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 21034
    },
    {
      "epoch": 8.403915301638035,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 21035
    },
    {
      "epoch": 8.404314822213344,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 21036
    },
    {
      "epoch": 8.404714342788653,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 21037
    },
    {
      "epoch": 8.405113863363963,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 21038
    },
    {
      "epoch": 8.405513383939272,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 21039
    },
    {
      "epoch": 8.405912904514583,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 21040
    },
    {
      "epoch": 8.406312425089892,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 21041
    },
    {
      "epoch": 8.4067119456652,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 21042
    },
    {
      "epoch": 8.407111466240512,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 21043
    },
    {
      "epoch": 8.40751098681582,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 21044
    },
    {
      "epoch": 8.407910507391131,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 21045
    },
    {
      "epoch": 8.40831002796644,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 21046
    },
    {
      "epoch": 8.40870954854175,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 21047
    },
    {
      "epoch": 8.40910906911706,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 21048
    },
    {
      "epoch": 8.409508589692368,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 21049
    },
    {
      "epoch": 8.40990811026768,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 21050
    },
    {
      "epoch": 8.410307630842988,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 21051
    },
    {
      "epoch": 8.410707151418299,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 21052
    },
    {
      "epoch": 8.411106671993608,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 21053
    },
    {
      "epoch": 8.411506192568917,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 21054
    },
    {
      "epoch": 8.411905713144227,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 21055
    },
    {
      "epoch": 8.412305233719536,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 21056
    },
    {
      "epoch": 8.412704754294847,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 21057
    },
    {
      "epoch": 8.413104274870156,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 21058
    },
    {
      "epoch": 8.413503795445465,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 21059
    },
    {
      "epoch": 8.413903316020775,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 21060
    },
    {
      "epoch": 8.414302836596084,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 21061
    },
    {
      "epoch": 8.414702357171395,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 21062
    },
    {
      "epoch": 8.415101877746704,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 21063
    },
    {
      "epoch": 8.415501398322014,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 21064
    },
    {
      "epoch": 8.415900918897323,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 21065
    },
    {
      "epoch": 8.416300439472632,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 21066
    },
    {
      "epoch": 8.416699960047943,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 21067
    },
    {
      "epoch": 8.417099480623252,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 21068
    },
    {
      "epoch": 8.417499001198562,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 21069
    },
    {
      "epoch": 8.417898521773871,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 21070
    },
    {
      "epoch": 8.41829804234918,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 21071
    },
    {
      "epoch": 8.418697562924491,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 21072
    },
    {
      "epoch": 8.4190970834998,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 21073
    },
    {
      "epoch": 8.41949660407511,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 21074
    },
    {
      "epoch": 8.41989612465042,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 21075
    },
    {
      "epoch": 8.420295645225728,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 21076
    },
    {
      "epoch": 8.420695165801039,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 21077
    },
    {
      "epoch": 8.421094686376348,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 21078
    },
    {
      "epoch": 8.421494206951659,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 21079
    },
    {
      "epoch": 8.421893727526967,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 21080
    },
    {
      "epoch": 8.422293248102278,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 21081
    },
    {
      "epoch": 8.422692768677587,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 21082
    },
    {
      "epoch": 8.423092289252896,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 21083
    },
    {
      "epoch": 8.423491809828207,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 21084
    },
    {
      "epoch": 8.423891330403515,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 21085
    },
    {
      "epoch": 8.424290850978826,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 21086
    },
    {
      "epoch": 8.424690371554135,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 21087
    },
    {
      "epoch": 8.425089892129444,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 21088
    },
    {
      "epoch": 8.425489412704755,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 21089
    },
    {
      "epoch": 8.425888933280063,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 21090
    },
    {
      "epoch": 8.426288453855374,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 21091
    },
    {
      "epoch": 8.426687974430683,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 21092
    },
    {
      "epoch": 8.427087495005992,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 21093
    },
    {
      "epoch": 8.427487015581303,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 21094
    },
    {
      "epoch": 8.427886536156612,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 21095
    },
    {
      "epoch": 8.428286056731922,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 21096
    },
    {
      "epoch": 8.428685577307231,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 21097
    },
    {
      "epoch": 8.429085097882542,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 21098
    },
    {
      "epoch": 8.42948461845785,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 21099
    },
    {
      "epoch": 8.42988413903316,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 21100
    },
    {
      "epoch": 8.43028365960847,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 21101
    },
    {
      "epoch": 8.43068318018378,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 21102
    },
    {
      "epoch": 8.43108270075909,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 21103
    },
    {
      "epoch": 8.431482221334399,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 21104
    },
    {
      "epoch": 8.431881741909708,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 21105
    },
    {
      "epoch": 8.432281262485018,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 21106
    },
    {
      "epoch": 8.432680783060327,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 21107
    },
    {
      "epoch": 8.433080303635638,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 21108
    },
    {
      "epoch": 8.433479824210947,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 21109
    },
    {
      "epoch": 8.433879344786256,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 21110
    },
    {
      "epoch": 8.434278865361566,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 21111
    },
    {
      "epoch": 8.434678385936875,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 21112
    },
    {
      "epoch": 8.435077906512186,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 21113
    },
    {
      "epoch": 8.435477427087495,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 21114
    },
    {
      "epoch": 8.435876947662805,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 21115
    },
    {
      "epoch": 8.436276468238114,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 21116
    },
    {
      "epoch": 8.436675988813423,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 21117
    },
    {
      "epoch": 8.437075509388734,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 21118
    },
    {
      "epoch": 8.437475029964043,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 21119
    },
    {
      "epoch": 8.437874550539354,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 21120
    },
    {
      "epoch": 8.438274071114662,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 21121
    },
    {
      "epoch": 8.438673591689971,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 21122
    },
    {
      "epoch": 8.439073112265282,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 21123
    },
    {
      "epoch": 8.439472632840591,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 21124
    },
    {
      "epoch": 8.439872153415902,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 21125
    },
    {
      "epoch": 8.44027167399121,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 21126
    },
    {
      "epoch": 8.44067119456652,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 21127
    },
    {
      "epoch": 8.44107071514183,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 21128
    },
    {
      "epoch": 8.441470235717139,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 21129
    },
    {
      "epoch": 8.44186975629245,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 21130
    },
    {
      "epoch": 8.442269276867759,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 21131
    },
    {
      "epoch": 8.44266879744307,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 21132
    },
    {
      "epoch": 8.443068318018378,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 21133
    },
    {
      "epoch": 8.443467838593687,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 21134
    },
    {
      "epoch": 8.443867359168998,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 21135
    },
    {
      "epoch": 8.444266879744307,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 21136
    },
    {
      "epoch": 8.444666400319617,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 21137
    },
    {
      "epoch": 8.445065920894926,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 21138
    },
    {
      "epoch": 8.445465441470235,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 21139
    },
    {
      "epoch": 8.445864962045546,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 21140
    },
    {
      "epoch": 8.446264482620855,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 21141
    },
    {
      "epoch": 8.446664003196165,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 21142
    },
    {
      "epoch": 8.447063523771474,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 21143
    },
    {
      "epoch": 8.447463044346783,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 21144
    },
    {
      "epoch": 8.447862564922094,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 21145
    },
    {
      "epoch": 8.448262085497403,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 21146
    },
    {
      "epoch": 8.448661606072713,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 21147
    },
    {
      "epoch": 8.449061126648022,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 21148
    },
    {
      "epoch": 8.449460647223333,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 21149
    },
    {
      "epoch": 8.449860167798642,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 21150
    },
    {
      "epoch": 8.45025968837395,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 21151
    },
    {
      "epoch": 8.450659208949261,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 21152
    },
    {
      "epoch": 8.45105872952457,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 21153
    },
    {
      "epoch": 8.451458250099881,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 21154
    },
    {
      "epoch": 8.45185777067519,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 21155
    },
    {
      "epoch": 8.452257291250499,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 21156
    },
    {
      "epoch": 8.45265681182581,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 21157
    },
    {
      "epoch": 8.453056332401118,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 21158
    },
    {
      "epoch": 8.453455852976429,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 21159
    },
    {
      "epoch": 8.453855373551738,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0585,
      "step": 21160
    },
    {
      "epoch": 8.454254894127047,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 21161
    },
    {
      "epoch": 8.454654414702357,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 21162
    },
    {
      "epoch": 8.455053935277666,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 21163
    },
    {
      "epoch": 8.455453455852977,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 21164
    },
    {
      "epoch": 8.455852976428286,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 21165
    },
    {
      "epoch": 8.456252497003595,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 21166
    },
    {
      "epoch": 8.456652017578906,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 21167
    },
    {
      "epoch": 8.457051538154214,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 21168
    },
    {
      "epoch": 8.457451058729525,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 21169
    },
    {
      "epoch": 8.457850579304834,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 21170
    },
    {
      "epoch": 8.458250099880145,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 21171
    },
    {
      "epoch": 8.458649620455454,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 21172
    },
    {
      "epoch": 8.459049141030762,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 21173
    },
    {
      "epoch": 8.459448661606073,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 21174
    },
    {
      "epoch": 8.459848182181382,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 21175
    },
    {
      "epoch": 8.460247702756693,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 21176
    },
    {
      "epoch": 8.460647223332002,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 21177
    },
    {
      "epoch": 8.46104674390731,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 21178
    },
    {
      "epoch": 8.461446264482621,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 21179
    },
    {
      "epoch": 8.46184578505793,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 21180
    },
    {
      "epoch": 8.46224530563324,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 21181
    },
    {
      "epoch": 8.46264482620855,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 21182
    },
    {
      "epoch": 8.463044346783859,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 21183
    },
    {
      "epoch": 8.46344386735917,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 21184
    },
    {
      "epoch": 8.463843387934478,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 21185
    },
    {
      "epoch": 8.464242908509789,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 21186
    },
    {
      "epoch": 8.464642429085098,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 21187
    },
    {
      "epoch": 8.465041949660408,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 21188
    },
    {
      "epoch": 8.465441470235717,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 21189
    },
    {
      "epoch": 8.465840990811026,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 21190
    },
    {
      "epoch": 8.466240511386337,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 21191
    },
    {
      "epoch": 8.466640031961646,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 21192
    },
    {
      "epoch": 8.467039552536956,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 21193
    },
    {
      "epoch": 8.467439073112265,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 21194
    },
    {
      "epoch": 8.467838593687574,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 21195
    },
    {
      "epoch": 8.468238114262885,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 21196
    },
    {
      "epoch": 8.468637634838194,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 21197
    },
    {
      "epoch": 8.469037155413504,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 21198
    },
    {
      "epoch": 8.469436675988813,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 21199
    },
    {
      "epoch": 8.469836196564122,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 21200
    },
    {
      "epoch": 8.470235717139433,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 21201
    },
    {
      "epoch": 8.470635237714742,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 21202
    },
    {
      "epoch": 8.471034758290052,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 21203
    },
    {
      "epoch": 8.471434278865361,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 21204
    },
    {
      "epoch": 8.471833799440672,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 21205
    },
    {
      "epoch": 8.472233320015981,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 21206
    },
    {
      "epoch": 8.47263284059129,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 21207
    },
    {
      "epoch": 8.4730323611666,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 21208
    },
    {
      "epoch": 8.47343188174191,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 21209
    },
    {
      "epoch": 8.47383140231722,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 21210
    },
    {
      "epoch": 8.474230922892529,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 21211
    },
    {
      "epoch": 8.474630443467838,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 21212
    },
    {
      "epoch": 8.475029964043149,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 21213
    },
    {
      "epoch": 8.475429484618457,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 21214
    },
    {
      "epoch": 8.475829005193768,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 21215
    },
    {
      "epoch": 8.476228525769077,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 21216
    },
    {
      "epoch": 8.476628046344386,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 21217
    },
    {
      "epoch": 8.477027566919697,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 21218
    },
    {
      "epoch": 8.477427087495006,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 21219
    },
    {
      "epoch": 8.477826608070316,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 21220
    },
    {
      "epoch": 8.478226128645625,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 21221
    },
    {
      "epoch": 8.478625649220934,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 21222
    },
    {
      "epoch": 8.479025169796245,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 21223
    },
    {
      "epoch": 8.479424690371554,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 21224
    },
    {
      "epoch": 8.479824210946864,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 21225
    },
    {
      "epoch": 8.480223731522173,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 21226
    },
    {
      "epoch": 8.480623252097484,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 21227
    },
    {
      "epoch": 8.481022772672793,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 21228
    },
    {
      "epoch": 8.481422293248102,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 21229
    },
    {
      "epoch": 8.481821813823412,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 21230
    },
    {
      "epoch": 8.482221334398721,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 21231
    },
    {
      "epoch": 8.482620854974032,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 21232
    },
    {
      "epoch": 8.48302037554934,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 21233
    },
    {
      "epoch": 8.48341989612465,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 21234
    },
    {
      "epoch": 8.48381941669996,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 21235
    },
    {
      "epoch": 8.48421893727527,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 21236
    },
    {
      "epoch": 8.48461845785058,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0547,
      "step": 21237
    },
    {
      "epoch": 8.485017978425889,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 21238
    },
    {
      "epoch": 8.485417499001198,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 21239
    },
    {
      "epoch": 8.485817019576508,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 21240
    },
    {
      "epoch": 8.486216540151817,
      "grad_norm": 0.171875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 21241
    },
    {
      "epoch": 8.486616060727128,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 21242
    },
    {
      "epoch": 8.487015581302437,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 21243
    },
    {
      "epoch": 8.487415101877748,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 21244
    },
    {
      "epoch": 8.487814622453056,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 21245
    },
    {
      "epoch": 8.488214143028365,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 21246
    },
    {
      "epoch": 8.488613663603676,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 21247
    },
    {
      "epoch": 8.489013184178985,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 21248
    },
    {
      "epoch": 8.489412704754296,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 21249
    },
    {
      "epoch": 8.489812225329604,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 21250
    },
    {
      "epoch": 8.490211745904913,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 21251
    },
    {
      "epoch": 8.490611266480224,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 21252
    },
    {
      "epoch": 8.491010787055533,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 21253
    },
    {
      "epoch": 8.491410307630844,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 21254
    },
    {
      "epoch": 8.491809828206152,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 21255
    },
    {
      "epoch": 8.492209348781461,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 21256
    },
    {
      "epoch": 8.492608869356772,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 21257
    },
    {
      "epoch": 8.493008389932081,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 21258
    },
    {
      "epoch": 8.493407910507392,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 21259
    },
    {
      "epoch": 8.4938074310827,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 21260
    },
    {
      "epoch": 8.494206951658011,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 21261
    },
    {
      "epoch": 8.49460647223332,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 21262
    },
    {
      "epoch": 8.495005992808629,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 21263
    },
    {
      "epoch": 8.49540551338394,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 21264
    },
    {
      "epoch": 8.495805033959249,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 21265
    },
    {
      "epoch": 8.49620455453456,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 21266
    },
    {
      "epoch": 8.496604075109868,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 21267
    },
    {
      "epoch": 8.497003595685177,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 21268
    },
    {
      "epoch": 8.497403116260488,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 21269
    },
    {
      "epoch": 8.497802636835797,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 21270
    },
    {
      "epoch": 8.498202157411107,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 21271
    },
    {
      "epoch": 8.498601677986416,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 21272
    },
    {
      "epoch": 8.499001198561725,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 21273
    },
    {
      "epoch": 8.499400719137036,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 21274
    },
    {
      "epoch": 8.499800239712345,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 21275
    },
    {
      "epoch": 8.500199760287655,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 21276
    },
    {
      "epoch": 8.500599280862964,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 21277
    },
    {
      "epoch": 8.500998801438275,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 21278
    },
    {
      "epoch": 8.501398322013584,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 21279
    },
    {
      "epoch": 8.501797842588893,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 21280
    },
    {
      "epoch": 8.502197363164203,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 21281
    },
    {
      "epoch": 8.502596883739512,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 21282
    },
    {
      "epoch": 8.502996404314823,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 21283
    },
    {
      "epoch": 8.503395924890132,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 21284
    },
    {
      "epoch": 8.50379544546544,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 21285
    },
    {
      "epoch": 8.504194966040751,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 21286
    },
    {
      "epoch": 8.50459448661606,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 21287
    },
    {
      "epoch": 8.504994007191371,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 21288
    },
    {
      "epoch": 8.50539352776668,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 21289
    },
    {
      "epoch": 8.505793048341989,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 21290
    },
    {
      "epoch": 8.5061925689173,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 21291
    },
    {
      "epoch": 8.506592089492608,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 21292
    },
    {
      "epoch": 8.506991610067919,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 21293
    },
    {
      "epoch": 8.507391130643228,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 21294
    },
    {
      "epoch": 8.507790651218539,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 21295
    },
    {
      "epoch": 8.508190171793848,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 21296
    },
    {
      "epoch": 8.508589692369156,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 21297
    },
    {
      "epoch": 8.508989212944467,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 21298
    },
    {
      "epoch": 8.509388733519776,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 21299
    },
    {
      "epoch": 8.509788254095087,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 21300
    },
    {
      "epoch": 8.510187774670396,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 21301
    },
    {
      "epoch": 8.510587295245704,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 21302
    },
    {
      "epoch": 8.510986815821015,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 21303
    },
    {
      "epoch": 8.511386336396324,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 21304
    },
    {
      "epoch": 8.511785856971635,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 21305
    },
    {
      "epoch": 8.512185377546944,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 21306
    },
    {
      "epoch": 8.512584898122252,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 21307
    },
    {
      "epoch": 8.512984418697563,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 21308
    },
    {
      "epoch": 8.513383939272872,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 21309
    },
    {
      "epoch": 8.513783459848183,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 21310
    },
    {
      "epoch": 8.514182980423492,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 21311
    },
    {
      "epoch": 8.514582500998802,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 21312
    },
    {
      "epoch": 8.514982021574111,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 21313
    },
    {
      "epoch": 8.51538154214942,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 21314
    },
    {
      "epoch": 8.51578106272473,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 21315
    },
    {
      "epoch": 8.51618058330004,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 21316
    },
    {
      "epoch": 8.51658010387535,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 21317
    },
    {
      "epoch": 8.51697962445066,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 21318
    },
    {
      "epoch": 8.517379145025968,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 21319
    },
    {
      "epoch": 8.517778665601279,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 21320
    },
    {
      "epoch": 8.518178186176588,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 21321
    },
    {
      "epoch": 8.518577706751898,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 21322
    },
    {
      "epoch": 8.518977227327207,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 21323
    },
    {
      "epoch": 8.519376747902516,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 21324
    },
    {
      "epoch": 8.519776268477827,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 21325
    },
    {
      "epoch": 8.520175789053136,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 21326
    },
    {
      "epoch": 8.520575309628446,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 21327
    },
    {
      "epoch": 8.520974830203755,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 21328
    },
    {
      "epoch": 8.521374350779066,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 21329
    },
    {
      "epoch": 8.521773871354375,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 21330
    },
    {
      "epoch": 8.522173391929684,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 21331
    },
    {
      "epoch": 8.522572912504994,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 21332
    },
    {
      "epoch": 8.522972433080303,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 21333
    },
    {
      "epoch": 8.523371953655614,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 21334
    },
    {
      "epoch": 8.523771474230923,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 21335
    },
    {
      "epoch": 8.524170994806232,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 21336
    },
    {
      "epoch": 8.524570515381543,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 21337
    },
    {
      "epoch": 8.524970035956851,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 21338
    },
    {
      "epoch": 8.525369556532162,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 21339
    },
    {
      "epoch": 8.525769077107471,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 21340
    },
    {
      "epoch": 8.52616859768278,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 21341
    },
    {
      "epoch": 8.52656811825809,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 21342
    },
    {
      "epoch": 8.5269676388334,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 21343
    },
    {
      "epoch": 8.52736715940871,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 21344
    },
    {
      "epoch": 8.527766679984019,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 21345
    },
    {
      "epoch": 8.52816620055933,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 21346
    },
    {
      "epoch": 8.528565721134639,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 21347
    },
    {
      "epoch": 8.528965241709948,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 21348
    },
    {
      "epoch": 8.529364762285258,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 21349
    },
    {
      "epoch": 8.529764282860567,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 21350
    },
    {
      "epoch": 8.530163803435878,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 21351
    },
    {
      "epoch": 8.530563324011187,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 21352
    },
    {
      "epoch": 8.530962844586496,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 21353
    },
    {
      "epoch": 8.531362365161806,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 21354
    },
    {
      "epoch": 8.531761885737115,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 21355
    },
    {
      "epoch": 8.532161406312426,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 21356
    },
    {
      "epoch": 8.532560926887735,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 21357
    },
    {
      "epoch": 8.532960447463044,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 21358
    },
    {
      "epoch": 8.533359968038354,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 21359
    },
    {
      "epoch": 8.533759488613663,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 21360
    },
    {
      "epoch": 8.534159009188974,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 21361
    },
    {
      "epoch": 8.534558529764283,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 21362
    },
    {
      "epoch": 8.534958050339593,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 21363
    },
    {
      "epoch": 8.535357570914902,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 21364
    },
    {
      "epoch": 8.535757091490211,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 21365
    },
    {
      "epoch": 8.536156612065522,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 21366
    },
    {
      "epoch": 8.53655613264083,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 21367
    },
    {
      "epoch": 8.53695565321614,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 21368
    },
    {
      "epoch": 8.53735517379145,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 21369
    },
    {
      "epoch": 8.53775469436676,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 21370
    },
    {
      "epoch": 8.53815421494207,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 21371
    },
    {
      "epoch": 8.538553735517379,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 21372
    },
    {
      "epoch": 8.53895325609269,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 21373
    },
    {
      "epoch": 8.539352776667998,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 21374
    },
    {
      "epoch": 8.539752297243307,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 21375
    },
    {
      "epoch": 8.540151817818618,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 21376
    },
    {
      "epoch": 8.540551338393927,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 21377
    },
    {
      "epoch": 8.540950858969238,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 21378
    },
    {
      "epoch": 8.541350379544546,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 21379
    },
    {
      "epoch": 8.541749900119855,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 21380
    },
    {
      "epoch": 8.542149420695166,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 21381
    },
    {
      "epoch": 8.542548941270475,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 21382
    },
    {
      "epoch": 8.542948461845786,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 21383
    },
    {
      "epoch": 8.543347982421094,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 21384
    },
    {
      "epoch": 8.543747502996403,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 21385
    },
    {
      "epoch": 8.544147023571714,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 21386
    },
    {
      "epoch": 8.544546544147023,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 21387
    },
    {
      "epoch": 8.544946064722334,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 21388
    },
    {
      "epoch": 8.545345585297643,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 21389
    },
    {
      "epoch": 8.545745105872953,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 21390
    },
    {
      "epoch": 8.546144626448262,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 21391
    },
    {
      "epoch": 8.546544147023571,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 21392
    },
    {
      "epoch": 8.546943667598882,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 21393
    },
    {
      "epoch": 8.54734318817419,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 21394
    },
    {
      "epoch": 8.547742708749501,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 21395
    },
    {
      "epoch": 8.54814222932481,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 21396
    },
    {
      "epoch": 8.548541749900119,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 21397
    },
    {
      "epoch": 8.54894127047543,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 21398
    },
    {
      "epoch": 8.549340791050739,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 21399
    },
    {
      "epoch": 8.54974031162605,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 21400
    },
    {
      "epoch": 8.550139832201358,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 21401
    },
    {
      "epoch": 8.550539352776667,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 21402
    },
    {
      "epoch": 8.550938873351978,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 21403
    },
    {
      "epoch": 8.551338393927287,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 21404
    },
    {
      "epoch": 8.551737914502597,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 21405
    },
    {
      "epoch": 8.552137435077906,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 21406
    },
    {
      "epoch": 8.552536955653217,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 21407
    },
    {
      "epoch": 8.552936476228526,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 21408
    },
    {
      "epoch": 8.553335996803835,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 21409
    },
    {
      "epoch": 8.553735517379145,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 21410
    },
    {
      "epoch": 8.554135037954454,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 21411
    },
    {
      "epoch": 8.554534558529765,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 21412
    },
    {
      "epoch": 8.554934079105074,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 21413
    },
    {
      "epoch": 8.555333599680383,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 21414
    },
    {
      "epoch": 8.555733120255693,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 21415
    },
    {
      "epoch": 8.556132640831002,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 21416
    },
    {
      "epoch": 8.556532161406313,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 21417
    },
    {
      "epoch": 8.556931681981622,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 21418
    },
    {
      "epoch": 8.55733120255693,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 21419
    },
    {
      "epoch": 8.557730723132241,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 21420
    },
    {
      "epoch": 8.55813024370755,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 21421
    },
    {
      "epoch": 8.558529764282861,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 21422
    },
    {
      "epoch": 8.55892928485817,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 21423
    },
    {
      "epoch": 8.55932880543348,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 21424
    },
    {
      "epoch": 8.55972832600879,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 21425
    },
    {
      "epoch": 8.560127846584098,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 21426
    },
    {
      "epoch": 8.560527367159409,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 21427
    },
    {
      "epoch": 8.560926887734718,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 21428
    },
    {
      "epoch": 8.561326408310029,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 21429
    },
    {
      "epoch": 8.561725928885338,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 21430
    },
    {
      "epoch": 8.562125449460646,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 21431
    },
    {
      "epoch": 8.562524970035957,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 21432
    },
    {
      "epoch": 8.562924490611266,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 21433
    },
    {
      "epoch": 8.563324011186577,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 21434
    },
    {
      "epoch": 8.563723531761886,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 21435
    },
    {
      "epoch": 8.564123052337195,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 21436
    },
    {
      "epoch": 8.564522572912505,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 21437
    },
    {
      "epoch": 8.564922093487814,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 21438
    },
    {
      "epoch": 8.565321614063125,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 21439
    },
    {
      "epoch": 8.565721134638434,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 21440
    },
    {
      "epoch": 8.566120655213744,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 21441
    },
    {
      "epoch": 8.566520175789053,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 21442
    },
    {
      "epoch": 8.566919696364362,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 21443
    },
    {
      "epoch": 8.567319216939673,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 21444
    },
    {
      "epoch": 8.567718737514982,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 21445
    },
    {
      "epoch": 8.568118258090292,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 21446
    },
    {
      "epoch": 8.568517778665601,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 21447
    },
    {
      "epoch": 8.56891729924091,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 21448
    },
    {
      "epoch": 8.56931681981622,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 21449
    },
    {
      "epoch": 8.56971634039153,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 21450
    },
    {
      "epoch": 8.57011586096684,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 21451
    },
    {
      "epoch": 8.57051538154215,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 21452
    },
    {
      "epoch": 8.570914902117458,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 21453
    },
    {
      "epoch": 8.571314422692769,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 21454
    },
    {
      "epoch": 8.571713943268078,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 21455
    },
    {
      "epoch": 8.572113463843388,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 21456
    },
    {
      "epoch": 8.572512984418697,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 21457
    },
    {
      "epoch": 8.572912504994008,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 21458
    },
    {
      "epoch": 8.573312025569317,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 21459
    },
    {
      "epoch": 8.573711546144626,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 21460
    },
    {
      "epoch": 8.574111066719937,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 21461
    },
    {
      "epoch": 8.574510587295245,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 21462
    },
    {
      "epoch": 8.574910107870556,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 21463
    },
    {
      "epoch": 8.575309628445865,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 21464
    },
    {
      "epoch": 8.575709149021174,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 21465
    },
    {
      "epoch": 8.576108669596485,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 21466
    },
    {
      "epoch": 8.576508190171793,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 21467
    },
    {
      "epoch": 8.576907710747104,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 21468
    },
    {
      "epoch": 8.577307231322413,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 21469
    },
    {
      "epoch": 8.577706751897722,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 21470
    },
    {
      "epoch": 8.578106272473033,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 21471
    },
    {
      "epoch": 8.578505793048341,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 21472
    },
    {
      "epoch": 8.578905313623652,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 21473
    },
    {
      "epoch": 8.579304834198961,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 21474
    },
    {
      "epoch": 8.579704354774272,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 21475
    },
    {
      "epoch": 8.58010387534958,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 21476
    },
    {
      "epoch": 8.58050339592489,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 21477
    },
    {
      "epoch": 8.5809029165002,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 21478
    },
    {
      "epoch": 8.581302437075509,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 21479
    },
    {
      "epoch": 8.58170195765082,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 21480
    },
    {
      "epoch": 8.582101478226129,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 21481
    },
    {
      "epoch": 8.582500998801438,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 21482
    },
    {
      "epoch": 8.582900519376748,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 21483
    },
    {
      "epoch": 8.583300039952057,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 21484
    },
    {
      "epoch": 8.583699560527368,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 21485
    },
    {
      "epoch": 8.584099081102677,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 21486
    },
    {
      "epoch": 8.584498601677986,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 21487
    },
    {
      "epoch": 8.584898122253296,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 21488
    },
    {
      "epoch": 8.585297642828605,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 21489
    },
    {
      "epoch": 8.585697163403916,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 21490
    },
    {
      "epoch": 8.586096683979225,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 21491
    },
    {
      "epoch": 8.586496204554535,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 21492
    },
    {
      "epoch": 8.586895725129844,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 21493
    },
    {
      "epoch": 8.587295245705153,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 21494
    },
    {
      "epoch": 8.587694766280464,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 21495
    },
    {
      "epoch": 8.588094286855773,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 21496
    },
    {
      "epoch": 8.588493807431083,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 21497
    },
    {
      "epoch": 8.588893328006392,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 21498
    },
    {
      "epoch": 8.589292848581701,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 21499
    },
    {
      "epoch": 8.589692369157012,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 21500
    },
    {
      "epoch": 8.59009188973232,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 21501
    },
    {
      "epoch": 8.590491410307632,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 21502
    },
    {
      "epoch": 8.59089093088294,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 21503
    },
    {
      "epoch": 8.59129045145825,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 21504
    },
    {
      "epoch": 8.59168997203356,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 21505
    },
    {
      "epoch": 8.592089492608869,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 21506
    },
    {
      "epoch": 8.59248901318418,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 21507
    },
    {
      "epoch": 8.592888533759488,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 21508
    },
    {
      "epoch": 8.5932880543348,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 21509
    },
    {
      "epoch": 8.593687574910108,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 21510
    },
    {
      "epoch": 8.594087095485417,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 21511
    },
    {
      "epoch": 8.594486616060728,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 21512
    },
    {
      "epoch": 8.594886136636037,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 21513
    },
    {
      "epoch": 8.595285657211347,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 21514
    },
    {
      "epoch": 8.595685177786656,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 21515
    },
    {
      "epoch": 8.596084698361965,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 21516
    },
    {
      "epoch": 8.596484218937276,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 21517
    },
    {
      "epoch": 8.596883739512585,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 21518
    },
    {
      "epoch": 8.597283260087895,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 21519
    },
    {
      "epoch": 8.597682780663204,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 21520
    },
    {
      "epoch": 8.598082301238513,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 21521
    },
    {
      "epoch": 8.598481821813824,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 21522
    },
    {
      "epoch": 8.598881342389133,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 21523
    },
    {
      "epoch": 8.599280862964443,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 21524
    },
    {
      "epoch": 8.599680383539752,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 21525
    },
    {
      "epoch": 8.600079904115063,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 21526
    },
    {
      "epoch": 8.600479424690372,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 21527
    },
    {
      "epoch": 8.60087894526568,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 21528
    },
    {
      "epoch": 8.601278465840991,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 21529
    },
    {
      "epoch": 8.6016779864163,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 21530
    },
    {
      "epoch": 8.602077506991611,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 21531
    },
    {
      "epoch": 8.60247702756692,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 21532
    },
    {
      "epoch": 8.602876548142229,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 21533
    },
    {
      "epoch": 8.60327606871754,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0542,
      "step": 21534
    },
    {
      "epoch": 8.603675589292848,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 21535
    },
    {
      "epoch": 8.604075109868159,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 21536
    },
    {
      "epoch": 8.604474630443468,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 21537
    },
    {
      "epoch": 8.604874151018777,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 21538
    },
    {
      "epoch": 8.605273671594087,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 21539
    },
    {
      "epoch": 8.605673192169396,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 21540
    },
    {
      "epoch": 8.606072712744707,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 21541
    },
    {
      "epoch": 8.606472233320016,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 21542
    },
    {
      "epoch": 8.606871753895327,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 21543
    },
    {
      "epoch": 8.607271274470635,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 21544
    },
    {
      "epoch": 8.607670795045944,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 21545
    },
    {
      "epoch": 8.608070315621255,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 21546
    },
    {
      "epoch": 8.608469836196564,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 21547
    },
    {
      "epoch": 8.608869356771873,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 21548
    },
    {
      "epoch": 8.609268877347183,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 21549
    },
    {
      "epoch": 8.609668397922492,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 21550
    },
    {
      "epoch": 8.610067918497803,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 21551
    },
    {
      "epoch": 8.610467439073112,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 21552
    },
    {
      "epoch": 8.610866959648423,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 21553
    },
    {
      "epoch": 8.611266480223732,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 21554
    },
    {
      "epoch": 8.61166600079904,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 21555
    },
    {
      "epoch": 8.612065521374351,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 21556
    },
    {
      "epoch": 8.61246504194966,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 21557
    },
    {
      "epoch": 8.61286456252497,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 21558
    },
    {
      "epoch": 8.61326408310028,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 21559
    },
    {
      "epoch": 8.613663603675588,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 21560
    },
    {
      "epoch": 8.6140631242509,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 21561
    },
    {
      "epoch": 8.614462644826208,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 21562
    },
    {
      "epoch": 8.614862165401519,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 21563
    },
    {
      "epoch": 8.615261685976828,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 21564
    },
    {
      "epoch": 8.615661206552137,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 21565
    },
    {
      "epoch": 8.616060727127447,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 21566
    },
    {
      "epoch": 8.616460247702756,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 21567
    },
    {
      "epoch": 8.616859768278067,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 21568
    },
    {
      "epoch": 8.617259288853376,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 21569
    },
    {
      "epoch": 8.617658809428686,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 21570
    },
    {
      "epoch": 8.618058330003995,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 21571
    },
    {
      "epoch": 8.618457850579304,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 21572
    },
    {
      "epoch": 8.618857371154615,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 21573
    },
    {
      "epoch": 8.619256891729924,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 21574
    },
    {
      "epoch": 8.619656412305234,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 21575
    },
    {
      "epoch": 8.620055932880543,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 21576
    },
    {
      "epoch": 8.620455453455852,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 21577
    },
    {
      "epoch": 8.620854974031163,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 21578
    },
    {
      "epoch": 8.621254494606472,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 21579
    },
    {
      "epoch": 8.621654015181782,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 21580
    },
    {
      "epoch": 8.622053535757091,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 21581
    },
    {
      "epoch": 8.6224530563324,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 21582
    },
    {
      "epoch": 8.622852576907711,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 21583
    },
    {
      "epoch": 8.62325209748302,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 21584
    },
    {
      "epoch": 8.62365161805833,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 21585
    },
    {
      "epoch": 8.62405113863364,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 21586
    },
    {
      "epoch": 8.62445065920895,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 21587
    },
    {
      "epoch": 8.624850179784259,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 21588
    },
    {
      "epoch": 8.625249700359568,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 21589
    },
    {
      "epoch": 8.625649220934879,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 21590
    },
    {
      "epoch": 8.626048741510187,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 21591
    },
    {
      "epoch": 8.626448262085498,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 21592
    },
    {
      "epoch": 8.626847782660807,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 21593
    },
    {
      "epoch": 8.627247303236116,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 21594
    },
    {
      "epoch": 8.627646823811427,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 21595
    },
    {
      "epoch": 8.628046344386735,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0548,
      "step": 21596
    },
    {
      "epoch": 8.628445864962046,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 21597
    },
    {
      "epoch": 8.628845385537355,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 21598
    },
    {
      "epoch": 8.629244906112664,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 21599
    },
    {
      "epoch": 8.629644426687975,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 21600
    },
    {
      "epoch": 8.630043947263283,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 21601
    },
    {
      "epoch": 8.630443467838594,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 21602
    },
    {
      "epoch": 8.630842988413903,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 21603
    },
    {
      "epoch": 8.631242508989214,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 21604
    },
    {
      "epoch": 8.631642029564523,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 21605
    },
    {
      "epoch": 8.632041550139832,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 21606
    },
    {
      "epoch": 8.632441070715142,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 21607
    },
    {
      "epoch": 8.632840591290451,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 21608
    },
    {
      "epoch": 8.633240111865762,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 21609
    },
    {
      "epoch": 8.63363963244107,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 21610
    },
    {
      "epoch": 8.63403915301638,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 21611
    },
    {
      "epoch": 8.63443867359169,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 21612
    },
    {
      "epoch": 8.634838194167,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 21613
    },
    {
      "epoch": 8.63523771474231,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 21614
    },
    {
      "epoch": 8.635637235317619,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 21615
    },
    {
      "epoch": 8.636036755892928,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 21616
    },
    {
      "epoch": 8.636436276468238,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 21617
    },
    {
      "epoch": 8.636835797043547,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 21618
    },
    {
      "epoch": 8.637235317618858,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 21619
    },
    {
      "epoch": 8.637634838194167,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 21620
    },
    {
      "epoch": 8.638034358769477,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 21621
    },
    {
      "epoch": 8.638433879344786,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 21622
    },
    {
      "epoch": 8.638833399920095,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 21623
    },
    {
      "epoch": 8.639232920495406,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 21624
    },
    {
      "epoch": 8.639632441070715,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 21625
    },
    {
      "epoch": 8.640031961646025,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 21626
    },
    {
      "epoch": 8.640431482221334,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 21627
    },
    {
      "epoch": 8.640831002796643,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 21628
    },
    {
      "epoch": 8.641230523371954,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 21629
    },
    {
      "epoch": 8.641630043947263,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 21630
    },
    {
      "epoch": 8.642029564522574,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 21631
    },
    {
      "epoch": 8.642429085097882,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 21632
    },
    {
      "epoch": 8.642828605673191,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 21633
    },
    {
      "epoch": 8.643228126248502,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 21634
    },
    {
      "epoch": 8.643627646823811,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 21635
    },
    {
      "epoch": 8.644027167399122,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 21636
    },
    {
      "epoch": 8.64442668797443,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 21637
    },
    {
      "epoch": 8.644826208549741,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 21638
    },
    {
      "epoch": 8.64522572912505,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 21639
    },
    {
      "epoch": 8.645625249700359,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 21640
    },
    {
      "epoch": 8.64602477027567,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 21641
    },
    {
      "epoch": 8.646424290850979,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 21642
    },
    {
      "epoch": 8.64682381142629,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 21643
    },
    {
      "epoch": 8.647223332001598,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 21644
    },
    {
      "epoch": 8.647622852576907,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 21645
    },
    {
      "epoch": 8.648022373152218,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 21646
    },
    {
      "epoch": 8.648421893727527,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 21647
    },
    {
      "epoch": 8.648821414302837,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 21648
    },
    {
      "epoch": 8.649220934878146,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 21649
    },
    {
      "epoch": 8.649620455453455,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 21650
    },
    {
      "epoch": 8.650019976028766,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 21651
    },
    {
      "epoch": 8.650419496604075,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 21652
    },
    {
      "epoch": 8.650819017179385,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 21653
    },
    {
      "epoch": 8.651218537754694,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 21654
    },
    {
      "epoch": 8.651618058330005,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 21655
    },
    {
      "epoch": 8.652017578905314,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 21656
    },
    {
      "epoch": 8.652417099480623,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 21657
    },
    {
      "epoch": 8.652816620055933,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 21658
    },
    {
      "epoch": 8.653216140631242,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 21659
    },
    {
      "epoch": 8.653615661206553,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 21660
    },
    {
      "epoch": 8.654015181781862,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 21661
    },
    {
      "epoch": 8.65441470235717,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 21662
    },
    {
      "epoch": 8.654814222932481,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 21663
    },
    {
      "epoch": 8.65521374350779,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 21664
    },
    {
      "epoch": 8.655613264083101,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 21665
    },
    {
      "epoch": 8.65601278465841,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 21666
    },
    {
      "epoch": 8.656412305233719,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 21667
    },
    {
      "epoch": 8.65681182580903,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 21668
    },
    {
      "epoch": 8.657211346384338,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 21669
    },
    {
      "epoch": 8.657610866959649,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 21670
    },
    {
      "epoch": 8.658010387534958,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 21671
    },
    {
      "epoch": 8.658409908110269,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 21672
    },
    {
      "epoch": 8.658809428685577,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 21673
    },
    {
      "epoch": 8.659208949260886,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 21674
    },
    {
      "epoch": 8.659608469836197,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 21675
    },
    {
      "epoch": 8.660007990411506,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 21676
    },
    {
      "epoch": 8.660407510986817,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 21677
    },
    {
      "epoch": 8.660807031562126,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 21678
    },
    {
      "epoch": 8.661206552137434,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 21679
    },
    {
      "epoch": 8.661606072712745,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 21680
    },
    {
      "epoch": 8.662005593288054,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 21681
    },
    {
      "epoch": 8.662405113863365,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 21682
    },
    {
      "epoch": 8.662804634438674,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 21683
    },
    {
      "epoch": 8.663204155013982,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 21684
    },
    {
      "epoch": 8.663603675589293,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 21685
    },
    {
      "epoch": 8.664003196164602,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 21686
    },
    {
      "epoch": 8.664402716739913,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 21687
    },
    {
      "epoch": 8.664802237315222,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 21688
    },
    {
      "epoch": 8.665201757890532,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 21689
    },
    {
      "epoch": 8.665601278465841,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 21690
    },
    {
      "epoch": 8.66600079904115,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 21691
    },
    {
      "epoch": 8.66640031961646,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 21692
    },
    {
      "epoch": 8.66679984019177,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 21693
    },
    {
      "epoch": 8.66719936076708,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 21694
    },
    {
      "epoch": 8.66759888134239,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 21695
    },
    {
      "epoch": 8.667998401917698,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 21696
    },
    {
      "epoch": 8.668397922493009,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 21697
    },
    {
      "epoch": 8.668797443068318,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 21698
    },
    {
      "epoch": 8.669196963643628,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 21699
    },
    {
      "epoch": 8.669596484218937,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 21700
    },
    {
      "epoch": 8.669996004794246,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 21701
    },
    {
      "epoch": 8.670395525369557,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 21702
    },
    {
      "epoch": 8.670795045944866,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 21703
    },
    {
      "epoch": 8.671194566520176,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 21704
    },
    {
      "epoch": 8.671594087095485,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 21705
    },
    {
      "epoch": 8.671993607670796,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 21706
    },
    {
      "epoch": 8.672393128246105,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 21707
    },
    {
      "epoch": 8.672792648821414,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 21708
    },
    {
      "epoch": 8.673192169396724,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 21709
    },
    {
      "epoch": 8.673591689972033,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 21710
    },
    {
      "epoch": 8.673991210547344,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 21711
    },
    {
      "epoch": 8.674390731122653,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 21712
    },
    {
      "epoch": 8.674790251697962,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 21713
    },
    {
      "epoch": 8.675189772273272,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 21714
    },
    {
      "epoch": 8.675589292848581,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 21715
    },
    {
      "epoch": 8.675988813423892,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 21716
    },
    {
      "epoch": 8.676388333999201,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 21717
    },
    {
      "epoch": 8.67678785457451,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 21718
    },
    {
      "epoch": 8.67718737514982,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 21719
    },
    {
      "epoch": 8.67758689572513,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 21720
    },
    {
      "epoch": 8.67798641630044,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 21721
    },
    {
      "epoch": 8.678385936875749,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 21722
    },
    {
      "epoch": 8.67878545745106,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 21723
    },
    {
      "epoch": 8.679184978026369,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 21724
    },
    {
      "epoch": 8.679584498601677,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 21725
    },
    {
      "epoch": 8.679984019176988,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 21726
    },
    {
      "epoch": 8.680383539752297,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 21727
    },
    {
      "epoch": 8.680783060327606,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 21728
    },
    {
      "epoch": 8.681182580902917,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 21729
    },
    {
      "epoch": 8.681582101478226,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 21730
    },
    {
      "epoch": 8.681981622053536,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 21731
    },
    {
      "epoch": 8.682381142628845,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 21732
    },
    {
      "epoch": 8.682780663204156,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 21733
    },
    {
      "epoch": 8.683180183779465,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 21734
    },
    {
      "epoch": 8.683579704354774,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 21735
    },
    {
      "epoch": 8.683979224930084,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 21736
    },
    {
      "epoch": 8.684378745505393,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 21737
    },
    {
      "epoch": 8.684778266080704,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 21738
    },
    {
      "epoch": 8.685177786656013,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 21739
    },
    {
      "epoch": 8.685577307231322,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 21740
    },
    {
      "epoch": 8.685976827806632,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 21741
    },
    {
      "epoch": 8.686376348381941,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 21742
    },
    {
      "epoch": 8.686775868957252,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 21743
    },
    {
      "epoch": 8.68717538953256,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 21744
    },
    {
      "epoch": 8.68757491010787,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 21745
    },
    {
      "epoch": 8.68797443068318,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 21746
    },
    {
      "epoch": 8.68837395125849,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 21747
    },
    {
      "epoch": 8.6887734718338,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 21748
    },
    {
      "epoch": 8.689172992409109,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 21749
    },
    {
      "epoch": 8.68957251298442,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 21750
    },
    {
      "epoch": 8.689972033559728,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 21751
    },
    {
      "epoch": 8.690371554135037,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 21752
    },
    {
      "epoch": 8.690771074710348,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 21753
    },
    {
      "epoch": 8.691170595285657,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 21754
    },
    {
      "epoch": 8.691570115860968,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 21755
    },
    {
      "epoch": 8.691969636436276,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 21756
    },
    {
      "epoch": 8.692369157011585,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 21757
    },
    {
      "epoch": 8.692768677586896,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 21758
    },
    {
      "epoch": 8.693168198162205,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 21759
    },
    {
      "epoch": 8.693567718737516,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 21760
    },
    {
      "epoch": 8.693967239312824,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 21761
    },
    {
      "epoch": 8.694366759888133,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 21762
    },
    {
      "epoch": 8.694766280463444,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 21763
    },
    {
      "epoch": 8.695165801038753,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 21764
    },
    {
      "epoch": 8.695565321614064,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 21765
    },
    {
      "epoch": 8.695964842189372,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 21766
    },
    {
      "epoch": 8.696364362764683,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 21767
    },
    {
      "epoch": 8.696763883339992,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 21768
    },
    {
      "epoch": 8.697163403915301,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 21769
    },
    {
      "epoch": 8.697562924490612,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 21770
    },
    {
      "epoch": 8.69796244506592,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 21771
    },
    {
      "epoch": 8.698361965641231,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 21772
    },
    {
      "epoch": 8.69876148621654,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 21773
    },
    {
      "epoch": 8.699161006791849,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 21774
    },
    {
      "epoch": 8.69956052736716,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 21775
    },
    {
      "epoch": 8.699960047942469,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 21776
    },
    {
      "epoch": 8.70035956851778,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 21777
    },
    {
      "epoch": 8.700759089093088,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 21778
    },
    {
      "epoch": 8.701158609668397,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 21779
    },
    {
      "epoch": 8.701558130243708,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 21780
    },
    {
      "epoch": 8.701957650819017,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 21781
    },
    {
      "epoch": 8.702357171394327,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 21782
    },
    {
      "epoch": 8.702756691969636,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 21783
    },
    {
      "epoch": 8.703156212544947,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 21784
    },
    {
      "epoch": 8.703555733120256,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 21785
    },
    {
      "epoch": 8.703955253695565,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 21786
    },
    {
      "epoch": 8.704354774270875,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 21787
    },
    {
      "epoch": 8.704754294846184,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 21788
    },
    {
      "epoch": 8.705153815421495,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 21789
    },
    {
      "epoch": 8.705553335996804,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 21790
    },
    {
      "epoch": 8.705952856572113,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 21791
    },
    {
      "epoch": 8.706352377147423,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 21792
    },
    {
      "epoch": 8.706751897722732,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 21793
    },
    {
      "epoch": 8.707151418298043,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 21794
    },
    {
      "epoch": 8.707550938873352,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 21795
    },
    {
      "epoch": 8.70795045944866,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 21796
    },
    {
      "epoch": 8.708349980023971,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 21797
    },
    {
      "epoch": 8.70874950059928,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 21798
    },
    {
      "epoch": 8.709149021174591,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 21799
    },
    {
      "epoch": 8.7095485417499,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 21800
    },
    {
      "epoch": 8.70994806232521,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 21801
    },
    {
      "epoch": 8.71034758290052,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 21802
    },
    {
      "epoch": 8.710747103475828,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 21803
    },
    {
      "epoch": 8.711146624051139,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 21804
    },
    {
      "epoch": 8.711546144626448,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 21805
    },
    {
      "epoch": 8.711945665201759,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 21806
    },
    {
      "epoch": 8.712345185777068,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 21807
    },
    {
      "epoch": 8.712744706352376,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 21808
    },
    {
      "epoch": 8.713144226927687,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 21809
    },
    {
      "epoch": 8.713543747502996,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 21810
    },
    {
      "epoch": 8.713943268078307,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 21811
    },
    {
      "epoch": 8.714342788653616,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 21812
    },
    {
      "epoch": 8.714742309228924,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 21813
    },
    {
      "epoch": 8.715141829804235,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 21814
    },
    {
      "epoch": 8.715541350379544,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 21815
    },
    {
      "epoch": 8.715940870954855,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 21816
    },
    {
      "epoch": 8.716340391530164,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 21817
    },
    {
      "epoch": 8.716739912105474,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 21818
    },
    {
      "epoch": 8.717139432680783,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 21819
    },
    {
      "epoch": 8.717538953256092,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 21820
    },
    {
      "epoch": 8.717938473831403,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 21821
    },
    {
      "epoch": 8.718337994406712,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 21822
    },
    {
      "epoch": 8.718737514982022,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 21823
    },
    {
      "epoch": 8.719137035557331,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 21824
    },
    {
      "epoch": 8.71953655613264,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 21825
    },
    {
      "epoch": 8.71993607670795,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 21826
    },
    {
      "epoch": 8.72033559728326,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 21827
    },
    {
      "epoch": 8.72073511785857,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 21828
    },
    {
      "epoch": 8.72113463843388,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 21829
    },
    {
      "epoch": 8.721534159009188,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 21830
    },
    {
      "epoch": 8.721933679584499,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 21831
    },
    {
      "epoch": 8.722333200159808,
      "grad_norm": 0.03662109375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 21832
    },
    {
      "epoch": 8.722732720735118,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 21833
    },
    {
      "epoch": 8.723132241310427,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 21834
    },
    {
      "epoch": 8.723531761885738,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 21835
    },
    {
      "epoch": 8.723931282461047,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 21836
    },
    {
      "epoch": 8.724330803036356,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 21837
    },
    {
      "epoch": 8.724730323611666,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 21838
    },
    {
      "epoch": 8.725129844186975,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 21839
    },
    {
      "epoch": 8.725529364762286,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 21840
    },
    {
      "epoch": 8.725928885337595,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 21841
    },
    {
      "epoch": 8.726328405912904,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0115,
      "step": 21842
    },
    {
      "epoch": 8.726727926488214,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 21843
    },
    {
      "epoch": 8.727127447063523,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 21844
    },
    {
      "epoch": 8.727526967638834,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 21845
    },
    {
      "epoch": 8.727926488214143,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 21846
    },
    {
      "epoch": 8.728326008789452,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 21847
    },
    {
      "epoch": 8.728725529364763,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 21848
    },
    {
      "epoch": 8.729125049940071,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 21849
    },
    {
      "epoch": 8.729524570515382,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 21850
    },
    {
      "epoch": 8.729924091090691,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 21851
    },
    {
      "epoch": 8.730323611666002,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 21852
    },
    {
      "epoch": 8.73072313224131,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 21853
    },
    {
      "epoch": 8.73112265281662,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 21854
    },
    {
      "epoch": 8.73152217339193,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 21855
    },
    {
      "epoch": 8.731921693967239,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 21856
    },
    {
      "epoch": 8.73232121454255,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 21857
    },
    {
      "epoch": 8.732720735117859,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 21858
    },
    {
      "epoch": 8.733120255693168,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 21859
    },
    {
      "epoch": 8.733519776268478,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 21860
    },
    {
      "epoch": 8.733919296843787,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 21861
    },
    {
      "epoch": 8.734318817419098,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 21862
    },
    {
      "epoch": 8.734718337994407,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 21863
    },
    {
      "epoch": 8.735117858569716,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 21864
    },
    {
      "epoch": 8.735517379145026,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 21865
    },
    {
      "epoch": 8.735916899720335,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 21866
    },
    {
      "epoch": 8.736316420295646,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 21867
    },
    {
      "epoch": 8.736715940870955,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 21868
    },
    {
      "epoch": 8.737115461446265,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 21869
    },
    {
      "epoch": 8.737514982021574,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 21870
    },
    {
      "epoch": 8.737914502596883,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 21871
    },
    {
      "epoch": 8.738314023172194,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 21872
    },
    {
      "epoch": 8.738713543747503,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 21873
    },
    {
      "epoch": 8.739113064322813,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 21874
    },
    {
      "epoch": 8.739512584898122,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 21875
    },
    {
      "epoch": 8.739912105473431,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 21876
    },
    {
      "epoch": 8.740311626048742,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 21877
    },
    {
      "epoch": 8.74071114662405,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 21878
    },
    {
      "epoch": 8.741110667199361,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 21879
    },
    {
      "epoch": 8.74151018777467,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 21880
    },
    {
      "epoch": 8.74190970834998,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 21881
    },
    {
      "epoch": 8.74230922892529,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 21882
    },
    {
      "epoch": 8.742708749500599,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0558,
      "step": 21883
    },
    {
      "epoch": 8.74310827007591,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 21884
    },
    {
      "epoch": 8.743507790651218,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 21885
    },
    {
      "epoch": 8.743907311226529,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 21886
    },
    {
      "epoch": 8.744306831801838,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 21887
    },
    {
      "epoch": 8.744706352377147,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 21888
    },
    {
      "epoch": 8.745105872952458,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 21889
    },
    {
      "epoch": 8.745505393527766,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 21890
    },
    {
      "epoch": 8.745904914103077,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 21891
    },
    {
      "epoch": 8.746304434678386,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 21892
    },
    {
      "epoch": 8.746703955253695,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 21893
    },
    {
      "epoch": 8.747103475829006,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 21894
    },
    {
      "epoch": 8.747502996404315,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 21895
    },
    {
      "epoch": 8.747902516979625,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 21896
    },
    {
      "epoch": 8.748302037554934,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 21897
    },
    {
      "epoch": 8.748701558130243,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 21898
    },
    {
      "epoch": 8.749101078705554,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 21899
    },
    {
      "epoch": 8.749500599280863,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 21900
    },
    {
      "epoch": 8.749900119856173,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 21901
    },
    {
      "epoch": 8.750299640431482,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 21902
    },
    {
      "epoch": 8.750699161006793,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 21903
    },
    {
      "epoch": 8.751098681582102,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 21904
    },
    {
      "epoch": 8.75149820215741,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 21905
    },
    {
      "epoch": 8.751897722732721,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 21906
    },
    {
      "epoch": 8.75229724330803,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 21907
    },
    {
      "epoch": 8.752696763883339,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 21908
    },
    {
      "epoch": 8.75309628445865,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 21909
    },
    {
      "epoch": 8.753495805033959,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 21910
    },
    {
      "epoch": 8.75389532560927,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 21911
    },
    {
      "epoch": 8.754294846184578,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 21912
    },
    {
      "epoch": 8.754694366759889,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 21913
    },
    {
      "epoch": 8.755093887335198,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 21914
    },
    {
      "epoch": 8.755493407910507,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 21915
    },
    {
      "epoch": 8.755892928485817,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 21916
    },
    {
      "epoch": 8.756292449061126,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 21917
    },
    {
      "epoch": 8.756691969636437,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 21918
    },
    {
      "epoch": 8.757091490211746,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 21919
    },
    {
      "epoch": 8.757491010787055,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 21920
    },
    {
      "epoch": 8.757890531362365,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 21921
    },
    {
      "epoch": 8.758290051937674,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 21922
    },
    {
      "epoch": 8.758689572512985,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 21923
    },
    {
      "epoch": 8.759089093088294,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 21924
    },
    {
      "epoch": 8.759488613663603,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 21925
    },
    {
      "epoch": 8.759888134238913,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 21926
    },
    {
      "epoch": 8.760287654814222,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 21927
    },
    {
      "epoch": 8.760687175389533,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 21928
    },
    {
      "epoch": 8.761086695964842,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 21929
    },
    {
      "epoch": 8.761486216540153,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 21930
    },
    {
      "epoch": 8.761885737115461,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 21931
    },
    {
      "epoch": 8.76228525769077,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 21932
    },
    {
      "epoch": 8.762684778266081,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 21933
    },
    {
      "epoch": 8.76308429884139,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 21934
    },
    {
      "epoch": 8.7634838194167,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 21935
    },
    {
      "epoch": 8.76388333999201,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 21936
    },
    {
      "epoch": 8.764282860567318,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 21937
    },
    {
      "epoch": 8.764682381142629,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 21938
    },
    {
      "epoch": 8.765081901717938,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 21939
    },
    {
      "epoch": 8.765481422293249,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 21940
    },
    {
      "epoch": 8.765880942868558,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 21941
    },
    {
      "epoch": 8.766280463443866,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 21942
    },
    {
      "epoch": 8.766679984019177,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 21943
    },
    {
      "epoch": 8.767079504594486,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 21944
    },
    {
      "epoch": 8.767479025169797,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 21945
    },
    {
      "epoch": 8.767878545745106,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 21946
    },
    {
      "epoch": 8.768278066320416,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 21947
    },
    {
      "epoch": 8.768677586895725,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 21948
    },
    {
      "epoch": 8.769077107471034,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 21949
    },
    {
      "epoch": 8.769476628046345,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 21950
    },
    {
      "epoch": 8.769876148621654,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 21951
    },
    {
      "epoch": 8.770275669196964,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 21952
    },
    {
      "epoch": 8.770675189772273,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 21953
    },
    {
      "epoch": 8.771074710347582,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 21954
    },
    {
      "epoch": 8.771474230922893,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 21955
    },
    {
      "epoch": 8.771873751498202,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 21956
    },
    {
      "epoch": 8.772273272073512,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 21957
    },
    {
      "epoch": 8.772672792648821,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 21958
    },
    {
      "epoch": 8.77307231322413,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 21959
    },
    {
      "epoch": 8.77347183379944,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 21960
    },
    {
      "epoch": 8.77387135437475,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 21961
    },
    {
      "epoch": 8.77427087495006,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 21962
    },
    {
      "epoch": 8.77467039552537,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 21963
    },
    {
      "epoch": 8.77506991610068,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 21964
    },
    {
      "epoch": 8.775469436675989,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 21965
    },
    {
      "epoch": 8.775868957251298,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 21966
    },
    {
      "epoch": 8.776268477826608,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 21967
    },
    {
      "epoch": 8.776667998401917,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 21968
    },
    {
      "epoch": 8.777067518977228,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 21969
    },
    {
      "epoch": 8.777467039552537,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 21970
    },
    {
      "epoch": 8.777866560127846,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 21971
    },
    {
      "epoch": 8.778266080703157,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 21972
    },
    {
      "epoch": 8.778665601278465,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 21973
    },
    {
      "epoch": 8.779065121853776,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 21974
    },
    {
      "epoch": 8.779464642429085,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 21975
    },
    {
      "epoch": 8.779864163004394,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 21976
    },
    {
      "epoch": 8.780263683579705,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 21977
    },
    {
      "epoch": 8.780663204155013,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 21978
    },
    {
      "epoch": 8.781062724730324,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 21979
    },
    {
      "epoch": 8.781462245305633,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 21980
    },
    {
      "epoch": 8.781861765880944,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 21981
    },
    {
      "epoch": 8.782261286456253,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 21982
    },
    {
      "epoch": 8.782660807031561,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 21983
    },
    {
      "epoch": 8.783060327606872,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 21984
    },
    {
      "epoch": 8.783459848182181,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 21985
    },
    {
      "epoch": 8.783859368757492,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 21986
    },
    {
      "epoch": 8.7842588893328,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0532,
      "step": 21987
    },
    {
      "epoch": 8.78465840990811,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 21988
    },
    {
      "epoch": 8.78505793048342,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 21989
    },
    {
      "epoch": 8.785457451058729,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 21990
    },
    {
      "epoch": 8.78585697163404,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 21991
    },
    {
      "epoch": 8.786256492209349,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 21992
    },
    {
      "epoch": 8.786656012784658,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 21993
    },
    {
      "epoch": 8.787055533359968,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 21994
    },
    {
      "epoch": 8.787455053935277,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 21995
    },
    {
      "epoch": 8.787854574510588,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 21996
    },
    {
      "epoch": 8.788254095085897,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 21997
    },
    {
      "epoch": 8.788653615661207,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 21998
    },
    {
      "epoch": 8.789053136236516,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 21999
    },
    {
      "epoch": 8.789452656811825,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 22000
    },
    {
      "epoch": 8.789852177387136,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 22001
    },
    {
      "epoch": 8.790251697962445,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 22002
    },
    {
      "epoch": 8.790651218537755,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 22003
    },
    {
      "epoch": 8.791050739113064,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 22004
    },
    {
      "epoch": 8.791450259688373,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 22005
    },
    {
      "epoch": 8.791849780263684,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 22006
    },
    {
      "epoch": 8.792249300838993,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 22007
    },
    {
      "epoch": 8.792648821414303,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 22008
    },
    {
      "epoch": 8.793048341989612,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 22009
    },
    {
      "epoch": 8.793447862564921,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 22010
    },
    {
      "epoch": 8.793847383140232,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 22011
    },
    {
      "epoch": 8.79424690371554,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 22012
    },
    {
      "epoch": 8.794646424290852,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 22013
    },
    {
      "epoch": 8.79504594486616,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 22014
    },
    {
      "epoch": 8.795445465441471,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 22015
    },
    {
      "epoch": 8.79584498601678,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 22016
    },
    {
      "epoch": 8.796244506592089,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 22017
    },
    {
      "epoch": 8.7966440271674,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 22018
    },
    {
      "epoch": 8.797043547742708,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 22019
    },
    {
      "epoch": 8.79744306831802,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 22020
    },
    {
      "epoch": 8.797842588893328,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 22021
    },
    {
      "epoch": 8.798242109468637,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 22022
    },
    {
      "epoch": 8.798641630043948,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 22023
    },
    {
      "epoch": 8.799041150619257,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 22024
    },
    {
      "epoch": 8.799440671194567,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 22025
    },
    {
      "epoch": 8.799840191769876,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 22026
    },
    {
      "epoch": 8.800239712345185,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 22027
    },
    {
      "epoch": 8.800639232920496,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 22028
    },
    {
      "epoch": 8.801038753495805,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 22029
    },
    {
      "epoch": 8.801438274071115,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 22030
    },
    {
      "epoch": 8.801837794646424,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 22031
    },
    {
      "epoch": 8.802237315221735,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 22032
    },
    {
      "epoch": 8.802636835797044,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 22033
    },
    {
      "epoch": 8.803036356372353,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 22034
    },
    {
      "epoch": 8.803435876947663,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 22035
    },
    {
      "epoch": 8.803835397522972,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 22036
    },
    {
      "epoch": 8.804234918098283,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 22037
    },
    {
      "epoch": 8.804634438673592,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 22038
    },
    {
      "epoch": 8.8050339592489,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 22039
    },
    {
      "epoch": 8.805433479824211,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 22040
    },
    {
      "epoch": 8.80583300039952,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 22041
    },
    {
      "epoch": 8.806232520974831,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 22042
    },
    {
      "epoch": 8.80663204155014,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 22043
    },
    {
      "epoch": 8.807031562125449,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 22044
    },
    {
      "epoch": 8.80743108270076,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 22045
    },
    {
      "epoch": 8.807830603276068,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 22046
    },
    {
      "epoch": 8.808230123851379,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 22047
    },
    {
      "epoch": 8.808629644426688,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 22048
    },
    {
      "epoch": 8.809029165001999,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 22049
    },
    {
      "epoch": 8.809428685577307,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 22050
    },
    {
      "epoch": 8.809828206152616,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 22051
    },
    {
      "epoch": 8.810227726727927,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 22052
    },
    {
      "epoch": 8.810627247303236,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 22053
    },
    {
      "epoch": 8.811026767878547,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 22054
    },
    {
      "epoch": 8.811426288453855,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 22055
    },
    {
      "epoch": 8.811825809029164,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 22056
    },
    {
      "epoch": 8.812225329604475,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 22057
    },
    {
      "epoch": 8.812624850179784,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 22058
    },
    {
      "epoch": 8.813024370755095,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 22059
    },
    {
      "epoch": 8.813423891330403,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 22060
    },
    {
      "epoch": 8.813823411905712,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 22061
    },
    {
      "epoch": 8.814222932481023,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 22062
    },
    {
      "epoch": 8.814622453056332,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 22063
    },
    {
      "epoch": 8.815021973631643,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 22064
    },
    {
      "epoch": 8.815421494206952,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 22065
    },
    {
      "epoch": 8.815821014782262,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 22066
    },
    {
      "epoch": 8.816220535357571,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 22067
    },
    {
      "epoch": 8.81662005593288,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 22068
    },
    {
      "epoch": 8.81701957650819,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 22069
    },
    {
      "epoch": 8.8174190970835,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 22070
    },
    {
      "epoch": 8.81781861765881,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 22071
    },
    {
      "epoch": 8.81821813823412,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 22072
    },
    {
      "epoch": 8.818617658809428,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 22073
    },
    {
      "epoch": 8.819017179384739,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 22074
    },
    {
      "epoch": 8.819416699960048,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 22075
    },
    {
      "epoch": 8.819816220535358,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 22076
    },
    {
      "epoch": 8.820215741110667,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 22077
    },
    {
      "epoch": 8.820615261685976,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 22078
    },
    {
      "epoch": 8.821014782261287,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 22079
    },
    {
      "epoch": 8.821414302836596,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 22080
    },
    {
      "epoch": 8.821813823411906,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 22081
    },
    {
      "epoch": 8.822213343987215,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 22082
    },
    {
      "epoch": 8.822612864562524,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 22083
    },
    {
      "epoch": 8.823012385137835,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 22084
    },
    {
      "epoch": 8.823411905713144,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 22085
    },
    {
      "epoch": 8.823811426288454,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 22086
    },
    {
      "epoch": 8.824210946863763,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 22087
    },
    {
      "epoch": 8.824610467439072,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 22088
    },
    {
      "epoch": 8.825009988014383,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 22089
    },
    {
      "epoch": 8.825409508589692,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 22090
    },
    {
      "epoch": 8.825809029165002,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 22091
    },
    {
      "epoch": 8.826208549740311,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 22092
    },
    {
      "epoch": 8.826608070315622,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 22093
    },
    {
      "epoch": 8.827007590890931,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 22094
    },
    {
      "epoch": 8.82740711146624,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 22095
    },
    {
      "epoch": 8.82780663204155,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 22096
    },
    {
      "epoch": 8.82820615261686,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 22097
    },
    {
      "epoch": 8.82860567319217,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 22098
    },
    {
      "epoch": 8.829005193767479,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 22099
    },
    {
      "epoch": 8.829404714342788,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 22100
    },
    {
      "epoch": 8.829804234918099,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 22101
    },
    {
      "epoch": 8.830203755493407,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 22102
    },
    {
      "epoch": 8.830603276068718,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 22103
    },
    {
      "epoch": 8.831002796644027,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 22104
    },
    {
      "epoch": 8.831402317219336,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 22105
    },
    {
      "epoch": 8.831801837794647,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 22106
    },
    {
      "epoch": 8.832201358369955,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 22107
    },
    {
      "epoch": 8.832600878945266,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 22108
    },
    {
      "epoch": 8.833000399520575,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 22109
    },
    {
      "epoch": 8.833399920095886,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 22110
    },
    {
      "epoch": 8.833799440671195,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 22111
    },
    {
      "epoch": 8.834198961246504,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 22112
    },
    {
      "epoch": 8.834598481821814,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 22113
    },
    {
      "epoch": 8.834998002397123,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 22114
    },
    {
      "epoch": 8.835397522972434,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 22115
    },
    {
      "epoch": 8.835797043547743,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 22116
    },
    {
      "epoch": 8.836196564123052,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 22117
    },
    {
      "epoch": 8.836596084698362,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 22118
    },
    {
      "epoch": 8.836995605273671,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 22119
    },
    {
      "epoch": 8.837395125848982,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 22120
    },
    {
      "epoch": 8.83779464642429,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 22121
    },
    {
      "epoch": 8.8381941669996,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 22122
    },
    {
      "epoch": 8.83859368757491,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 22123
    },
    {
      "epoch": 8.83899320815022,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 22124
    },
    {
      "epoch": 8.83939272872553,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 22125
    },
    {
      "epoch": 8.839792249300839,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 22126
    },
    {
      "epoch": 8.84019176987615,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 22127
    },
    {
      "epoch": 8.840591290451458,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 22128
    },
    {
      "epoch": 8.840990811026767,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 22129
    },
    {
      "epoch": 8.841390331602078,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 22130
    },
    {
      "epoch": 8.841789852177387,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 22131
    },
    {
      "epoch": 8.842189372752697,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 22132
    },
    {
      "epoch": 8.842588893328006,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 22133
    },
    {
      "epoch": 8.842988413903315,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 22134
    },
    {
      "epoch": 8.843387934478626,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 22135
    },
    {
      "epoch": 8.843787455053935,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 22136
    },
    {
      "epoch": 8.844186975629245,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 22137
    },
    {
      "epoch": 8.844586496204554,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 22138
    },
    {
      "epoch": 8.844986016779863,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 22139
    },
    {
      "epoch": 8.845385537355174,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 22140
    },
    {
      "epoch": 8.845785057930483,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 22141
    },
    {
      "epoch": 8.846184578505794,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 22142
    },
    {
      "epoch": 8.846584099081102,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 22143
    },
    {
      "epoch": 8.846983619656413,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 22144
    },
    {
      "epoch": 8.847383140231722,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 22145
    },
    {
      "epoch": 8.847782660807031,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 22146
    },
    {
      "epoch": 8.848182181382342,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 22147
    },
    {
      "epoch": 8.84858170195765,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 22148
    },
    {
      "epoch": 8.848981222532961,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 22149
    },
    {
      "epoch": 8.84938074310827,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 22150
    },
    {
      "epoch": 8.849780263683579,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 22151
    },
    {
      "epoch": 8.85017978425889,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 22152
    },
    {
      "epoch": 8.850579304834199,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 22153
    },
    {
      "epoch": 8.85097882540951,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 22154
    },
    {
      "epoch": 8.851378345984818,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 22155
    },
    {
      "epoch": 8.851777866560127,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 22156
    },
    {
      "epoch": 8.852177387135438,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 22157
    },
    {
      "epoch": 8.852576907710747,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 22158
    },
    {
      "epoch": 8.852976428286057,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 22159
    },
    {
      "epoch": 8.853375948861366,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 22160
    },
    {
      "epoch": 8.853775469436677,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 22161
    },
    {
      "epoch": 8.854174990011986,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 22162
    },
    {
      "epoch": 8.854574510587295,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 22163
    },
    {
      "epoch": 8.854974031162605,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 22164
    },
    {
      "epoch": 8.855373551737914,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 22165
    },
    {
      "epoch": 8.855773072313225,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 22166
    },
    {
      "epoch": 8.856172592888534,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 22167
    },
    {
      "epoch": 8.856572113463843,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 22168
    },
    {
      "epoch": 8.856971634039153,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 22169
    },
    {
      "epoch": 8.857371154614462,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 22170
    },
    {
      "epoch": 8.857770675189773,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 22171
    },
    {
      "epoch": 8.858170195765082,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 22172
    },
    {
      "epoch": 8.85856971634039,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 22173
    },
    {
      "epoch": 8.858969236915701,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 22174
    },
    {
      "epoch": 8.85936875749101,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 22175
    },
    {
      "epoch": 8.859768278066321,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 22176
    },
    {
      "epoch": 8.86016779864163,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 22177
    },
    {
      "epoch": 8.86056731921694,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 22178
    },
    {
      "epoch": 8.86096683979225,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 22179
    },
    {
      "epoch": 8.861366360367558,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 22180
    },
    {
      "epoch": 8.861765880942869,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 22181
    },
    {
      "epoch": 8.862165401518178,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 22182
    },
    {
      "epoch": 8.862564922093489,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 22183
    },
    {
      "epoch": 8.862964442668797,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 22184
    },
    {
      "epoch": 8.863363963244106,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 22185
    },
    {
      "epoch": 8.863763483819417,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 22186
    },
    {
      "epoch": 8.864163004394726,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 22187
    },
    {
      "epoch": 8.864562524970037,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 22188
    },
    {
      "epoch": 8.864962045545346,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 22189
    },
    {
      "epoch": 8.865361566120654,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 22190
    },
    {
      "epoch": 8.865761086695965,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 22191
    },
    {
      "epoch": 8.866160607271274,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 22192
    },
    {
      "epoch": 8.866560127846585,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 22193
    },
    {
      "epoch": 8.866959648421894,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 22194
    },
    {
      "epoch": 8.867359168997204,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 22195
    },
    {
      "epoch": 8.867758689572513,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 22196
    },
    {
      "epoch": 8.868158210147822,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 22197
    },
    {
      "epoch": 8.868557730723133,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 22198
    },
    {
      "epoch": 8.868957251298442,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 22199
    },
    {
      "epoch": 8.869356771873752,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 22200
    },
    {
      "epoch": 8.869756292449061,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 22201
    },
    {
      "epoch": 8.87015581302437,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 22202
    },
    {
      "epoch": 8.87055533359968,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 22203
    },
    {
      "epoch": 8.87095485417499,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 22204
    },
    {
      "epoch": 8.8713543747503,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 22205
    },
    {
      "epoch": 8.87175389532561,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 22206
    },
    {
      "epoch": 8.872153415900918,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 22207
    },
    {
      "epoch": 8.872552936476229,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 22208
    },
    {
      "epoch": 8.872952457051538,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 22209
    },
    {
      "epoch": 8.873351977626848,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 22210
    },
    {
      "epoch": 8.873751498202157,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 22211
    },
    {
      "epoch": 8.874151018777468,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 22212
    },
    {
      "epoch": 8.874550539352777,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 22213
    },
    {
      "epoch": 8.874950059928086,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 22214
    },
    {
      "epoch": 8.875349580503396,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 22215
    },
    {
      "epoch": 8.875749101078705,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 22216
    },
    {
      "epoch": 8.876148621654016,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 22217
    },
    {
      "epoch": 8.876548142229325,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 22218
    },
    {
      "epoch": 8.876947662804634,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 22219
    },
    {
      "epoch": 8.877347183379944,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 22220
    },
    {
      "epoch": 8.877746703955253,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 22221
    },
    {
      "epoch": 8.878146224530564,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 22222
    },
    {
      "epoch": 8.878545745105873,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 22223
    },
    {
      "epoch": 8.878945265681182,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 22224
    },
    {
      "epoch": 8.879344786256492,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 22225
    },
    {
      "epoch": 8.879744306831801,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 22226
    },
    {
      "epoch": 8.880143827407112,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 22227
    },
    {
      "epoch": 8.880543347982421,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 22228
    },
    {
      "epoch": 8.880942868557732,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 22229
    },
    {
      "epoch": 8.88134238913304,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 22230
    },
    {
      "epoch": 8.88174190970835,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 22231
    },
    {
      "epoch": 8.88214143028366,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 22232
    },
    {
      "epoch": 8.882540950858969,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 22233
    },
    {
      "epoch": 8.88294047143428,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 22234
    },
    {
      "epoch": 8.883339992009589,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 22235
    },
    {
      "epoch": 8.883739512584897,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 22236
    },
    {
      "epoch": 8.884139033160208,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 22237
    },
    {
      "epoch": 8.884538553735517,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 22238
    },
    {
      "epoch": 8.884938074310828,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 22239
    },
    {
      "epoch": 8.885337594886137,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 22240
    },
    {
      "epoch": 8.885737115461446,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 22241
    },
    {
      "epoch": 8.886136636036756,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 22242
    },
    {
      "epoch": 8.886536156612065,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 22243
    },
    {
      "epoch": 8.886935677187376,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 22244
    },
    {
      "epoch": 8.887335197762685,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 22245
    },
    {
      "epoch": 8.887734718337995,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 22246
    },
    {
      "epoch": 8.888134238913304,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 22247
    },
    {
      "epoch": 8.888533759488613,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0549,
      "step": 22248
    },
    {
      "epoch": 8.888933280063924,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 22249
    },
    {
      "epoch": 8.889332800639233,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 22250
    },
    {
      "epoch": 8.889732321214543,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 22251
    },
    {
      "epoch": 8.890131841789852,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 22252
    },
    {
      "epoch": 8.890531362365161,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 22253
    },
    {
      "epoch": 8.890930882940472,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 22254
    },
    {
      "epoch": 8.89133040351578,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 22255
    },
    {
      "epoch": 8.891729924091091,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 22256
    },
    {
      "epoch": 8.8921294446664,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 22257
    },
    {
      "epoch": 8.89252896524171,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 22258
    },
    {
      "epoch": 8.89292848581702,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 22259
    },
    {
      "epoch": 8.893328006392329,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 22260
    },
    {
      "epoch": 8.89372752696764,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 22261
    },
    {
      "epoch": 8.894127047542948,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 22262
    },
    {
      "epoch": 8.894526568118257,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 22263
    },
    {
      "epoch": 8.894926088693568,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 22264
    },
    {
      "epoch": 8.895325609268877,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 22265
    },
    {
      "epoch": 8.895725129844188,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 22266
    },
    {
      "epoch": 8.896124650419496,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 22267
    },
    {
      "epoch": 8.896524170994805,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 22268
    },
    {
      "epoch": 8.896923691570116,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 22269
    },
    {
      "epoch": 8.897323212145425,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 22270
    },
    {
      "epoch": 8.897722732720736,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 22271
    },
    {
      "epoch": 8.898122253296044,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 22272
    },
    {
      "epoch": 8.898521773871355,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 22273
    },
    {
      "epoch": 8.898921294446664,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 22274
    },
    {
      "epoch": 8.899320815021973,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 22275
    },
    {
      "epoch": 8.899720335597284,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 22276
    },
    {
      "epoch": 8.900119856172592,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 22277
    },
    {
      "epoch": 8.900519376747903,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 22278
    },
    {
      "epoch": 8.900918897323212,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 22279
    },
    {
      "epoch": 8.901318417898521,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 22280
    },
    {
      "epoch": 8.901717938473832,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 22281
    },
    {
      "epoch": 8.90211745904914,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 22282
    },
    {
      "epoch": 8.902516979624451,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 22283
    },
    {
      "epoch": 8.90291650019976,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 22284
    },
    {
      "epoch": 8.903316020775069,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 22285
    },
    {
      "epoch": 8.90371554135038,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 22286
    },
    {
      "epoch": 8.904115061925689,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 22287
    },
    {
      "epoch": 8.904514582501,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 22288
    },
    {
      "epoch": 8.904914103076308,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 22289
    },
    {
      "epoch": 8.905313623651619,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 22290
    },
    {
      "epoch": 8.905713144226928,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 22291
    },
    {
      "epoch": 8.906112664802237,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 22292
    },
    {
      "epoch": 8.906512185377547,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 22293
    },
    {
      "epoch": 8.906911705952856,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 22294
    },
    {
      "epoch": 8.907311226528167,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 22295
    },
    {
      "epoch": 8.907710747103476,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 22296
    },
    {
      "epoch": 8.908110267678785,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 22297
    },
    {
      "epoch": 8.908509788254095,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 22298
    },
    {
      "epoch": 8.908909308829404,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 22299
    },
    {
      "epoch": 8.909308829404715,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 22300
    },
    {
      "epoch": 8.909708349980024,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 22301
    },
    {
      "epoch": 8.910107870555333,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 22302
    },
    {
      "epoch": 8.910507391130643,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 22303
    },
    {
      "epoch": 8.910906911705952,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 22304
    },
    {
      "epoch": 8.911306432281263,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 22305
    },
    {
      "epoch": 8.911705952856572,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 22306
    },
    {
      "epoch": 8.912105473431883,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 22307
    },
    {
      "epoch": 8.912504994007191,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 22308
    },
    {
      "epoch": 8.9129045145825,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 22309
    },
    {
      "epoch": 8.913304035157811,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 22310
    },
    {
      "epoch": 8.91370355573312,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 22311
    },
    {
      "epoch": 8.91410307630843,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 22312
    },
    {
      "epoch": 8.91450259688374,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 22313
    },
    {
      "epoch": 8.914902117459048,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 22314
    },
    {
      "epoch": 8.915301638034359,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 22315
    },
    {
      "epoch": 8.915701158609668,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 22316
    },
    {
      "epoch": 8.916100679184979,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 22317
    },
    {
      "epoch": 8.916500199760288,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 22318
    },
    {
      "epoch": 8.916899720335596,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 22319
    },
    {
      "epoch": 8.917299240910907,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 22320
    },
    {
      "epoch": 8.917698761486216,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 22321
    },
    {
      "epoch": 8.918098282061527,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 22322
    },
    {
      "epoch": 8.918497802636836,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 22323
    },
    {
      "epoch": 8.918897323212146,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 22324
    },
    {
      "epoch": 8.919296843787455,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 22325
    },
    {
      "epoch": 8.919696364362764,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 22326
    },
    {
      "epoch": 8.920095884938075,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 22327
    },
    {
      "epoch": 8.920495405513384,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 22328
    },
    {
      "epoch": 8.920894926088694,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 22329
    },
    {
      "epoch": 8.921294446664003,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 22330
    },
    {
      "epoch": 8.921693967239312,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 22331
    },
    {
      "epoch": 8.922093487814623,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 22332
    },
    {
      "epoch": 8.922493008389932,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 22333
    },
    {
      "epoch": 8.922892528965242,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 22334
    },
    {
      "epoch": 8.923292049540551,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 22335
    },
    {
      "epoch": 8.92369157011586,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 22336
    },
    {
      "epoch": 8.92409109069117,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 22337
    },
    {
      "epoch": 8.92449061126648,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 22338
    },
    {
      "epoch": 8.92489013184179,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 22339
    },
    {
      "epoch": 8.9252896524171,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 22340
    },
    {
      "epoch": 8.92568917299241,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 22341
    },
    {
      "epoch": 8.926088693567719,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 22342
    },
    {
      "epoch": 8.926488214143028,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 22343
    },
    {
      "epoch": 8.926887734718338,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 22344
    },
    {
      "epoch": 8.927287255293647,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 22345
    },
    {
      "epoch": 8.927686775868958,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 22346
    },
    {
      "epoch": 8.928086296444267,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 22347
    },
    {
      "epoch": 8.928485817019576,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 22348
    },
    {
      "epoch": 8.928885337594886,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 22349
    },
    {
      "epoch": 8.929284858170195,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 22350
    },
    {
      "epoch": 8.929684378745506,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 22351
    },
    {
      "epoch": 8.930083899320815,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 22352
    },
    {
      "epoch": 8.930483419896124,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 22353
    },
    {
      "epoch": 8.930882940471434,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 22354
    },
    {
      "epoch": 8.931282461046743,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 22355
    },
    {
      "epoch": 8.931681981622054,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 22356
    },
    {
      "epoch": 8.932081502197363,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 22357
    },
    {
      "epoch": 8.932481022772674,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 22358
    },
    {
      "epoch": 8.932880543347983,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 22359
    },
    {
      "epoch": 8.933280063923291,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 22360
    },
    {
      "epoch": 8.933679584498602,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 22361
    },
    {
      "epoch": 8.934079105073911,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 22362
    },
    {
      "epoch": 8.934478625649222,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 22363
    },
    {
      "epoch": 8.93487814622453,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 22364
    },
    {
      "epoch": 8.93527766679984,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 22365
    },
    {
      "epoch": 8.93567718737515,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 22366
    },
    {
      "epoch": 8.936076707950459,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 22367
    },
    {
      "epoch": 8.93647622852577,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 22368
    },
    {
      "epoch": 8.936875749101079,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 22369
    },
    {
      "epoch": 8.937275269676388,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 22370
    },
    {
      "epoch": 8.937674790251698,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 22371
    },
    {
      "epoch": 8.938074310827007,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 22372
    },
    {
      "epoch": 8.938473831402318,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 22373
    },
    {
      "epoch": 8.938873351977627,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 22374
    },
    {
      "epoch": 8.939272872552937,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 22375
    },
    {
      "epoch": 8.939672393128246,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 22376
    },
    {
      "epoch": 8.940071913703555,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 22377
    },
    {
      "epoch": 8.940471434278866,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 22378
    },
    {
      "epoch": 8.940870954854175,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 22379
    },
    {
      "epoch": 8.941270475429485,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 22380
    },
    {
      "epoch": 8.941669996004794,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 22381
    },
    {
      "epoch": 8.942069516580103,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 22382
    },
    {
      "epoch": 8.942469037155414,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 22383
    },
    {
      "epoch": 8.942868557730723,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 22384
    },
    {
      "epoch": 8.943268078306033,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 22385
    },
    {
      "epoch": 8.943667598881342,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 22386
    },
    {
      "epoch": 8.944067119456651,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 22387
    },
    {
      "epoch": 8.944466640031962,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 22388
    },
    {
      "epoch": 8.94486616060727,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 22389
    },
    {
      "epoch": 8.945265681182581,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 22390
    },
    {
      "epoch": 8.94566520175789,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 22391
    },
    {
      "epoch": 8.946064722333201,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 22392
    },
    {
      "epoch": 8.94646424290851,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 22393
    },
    {
      "epoch": 8.946863763483819,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 22394
    },
    {
      "epoch": 8.94726328405913,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 22395
    },
    {
      "epoch": 8.947662804634438,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 22396
    },
    {
      "epoch": 8.948062325209749,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 22397
    },
    {
      "epoch": 8.948461845785058,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 22398
    },
    {
      "epoch": 8.948861366360367,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 22399
    },
    {
      "epoch": 8.949260886935678,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 22400
    },
    {
      "epoch": 8.949660407510986,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 22401
    },
    {
      "epoch": 8.950059928086297,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 22402
    },
    {
      "epoch": 8.950459448661606,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 22403
    },
    {
      "epoch": 8.950858969236915,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 22404
    },
    {
      "epoch": 8.951258489812226,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 22405
    },
    {
      "epoch": 8.951658010387535,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 22406
    },
    {
      "epoch": 8.952057530962845,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 22407
    },
    {
      "epoch": 8.952457051538154,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 22408
    },
    {
      "epoch": 8.952856572113465,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 22409
    },
    {
      "epoch": 8.953256092688774,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 22410
    },
    {
      "epoch": 8.953655613264083,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 22411
    },
    {
      "epoch": 8.954055133839393,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 22412
    },
    {
      "epoch": 8.954454654414702,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 22413
    },
    {
      "epoch": 8.954854174990013,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 22414
    },
    {
      "epoch": 8.955253695565322,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 22415
    },
    {
      "epoch": 8.95565321614063,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 22416
    },
    {
      "epoch": 8.956052736715941,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 22417
    },
    {
      "epoch": 8.95645225729125,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 22418
    },
    {
      "epoch": 8.95685177786656,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 22419
    },
    {
      "epoch": 8.95725129844187,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 22420
    },
    {
      "epoch": 8.957650819017179,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 22421
    },
    {
      "epoch": 8.95805033959249,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 22422
    },
    {
      "epoch": 8.958449860167798,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 22423
    },
    {
      "epoch": 8.958849380743109,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 22424
    },
    {
      "epoch": 8.959248901318418,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 22425
    },
    {
      "epoch": 8.959648421893728,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 22426
    },
    {
      "epoch": 8.960047942469037,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 22427
    },
    {
      "epoch": 8.960447463044346,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 22428
    },
    {
      "epoch": 8.960846983619657,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 22429
    },
    {
      "epoch": 8.961246504194966,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 22430
    },
    {
      "epoch": 8.961646024770276,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 22431
    },
    {
      "epoch": 8.962045545345585,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 22432
    },
    {
      "epoch": 8.962445065920894,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 22433
    },
    {
      "epoch": 8.962844586496205,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 22434
    },
    {
      "epoch": 8.963244107071514,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 22435
    },
    {
      "epoch": 8.963643627646825,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 22436
    },
    {
      "epoch": 8.964043148222133,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 22437
    },
    {
      "epoch": 8.964442668797442,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 22438
    },
    {
      "epoch": 8.964842189372753,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 22439
    },
    {
      "epoch": 8.965241709948062,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 22440
    },
    {
      "epoch": 8.965641230523373,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 22441
    },
    {
      "epoch": 8.966040751098681,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 22442
    },
    {
      "epoch": 8.96644027167399,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 22443
    },
    {
      "epoch": 8.966839792249301,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 22444
    },
    {
      "epoch": 8.96723931282461,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 22445
    },
    {
      "epoch": 8.96763883339992,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 22446
    },
    {
      "epoch": 8.96803835397523,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 22447
    },
    {
      "epoch": 8.968437874550538,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 22448
    },
    {
      "epoch": 8.968837395125849,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 22449
    },
    {
      "epoch": 8.969236915701158,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 22450
    },
    {
      "epoch": 8.969636436276469,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 22451
    },
    {
      "epoch": 8.970035956851778,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 22452
    },
    {
      "epoch": 8.970435477427088,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 22453
    },
    {
      "epoch": 8.970834998002397,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 22454
    },
    {
      "epoch": 8.971234518577706,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 22455
    },
    {
      "epoch": 8.971634039153017,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 22456
    },
    {
      "epoch": 8.972033559728326,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 22457
    },
    {
      "epoch": 8.972433080303636,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 22458
    },
    {
      "epoch": 8.972832600878945,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 22459
    },
    {
      "epoch": 8.973232121454254,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 22460
    },
    {
      "epoch": 8.973631642029565,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 22461
    },
    {
      "epoch": 8.974031162604874,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 22462
    },
    {
      "epoch": 8.974430683180184,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 22463
    },
    {
      "epoch": 8.974830203755493,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 22464
    },
    {
      "epoch": 8.975229724330802,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 22465
    },
    {
      "epoch": 8.975629244906113,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 22466
    },
    {
      "epoch": 8.976028765481422,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 22467
    },
    {
      "epoch": 8.976428286056732,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 22468
    },
    {
      "epoch": 8.976827806632041,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 22469
    },
    {
      "epoch": 8.977227327207352,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 22470
    },
    {
      "epoch": 8.97762684778266,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 22471
    },
    {
      "epoch": 8.97802636835797,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 22472
    },
    {
      "epoch": 8.97842588893328,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 22473
    },
    {
      "epoch": 8.97882540950859,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 22474
    },
    {
      "epoch": 8.9792249300839,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 22475
    },
    {
      "epoch": 8.979624450659209,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 22476
    },
    {
      "epoch": 8.980023971234518,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 22477
    },
    {
      "epoch": 8.980423491809828,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 22478
    },
    {
      "epoch": 8.980823012385137,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 22479
    },
    {
      "epoch": 8.981222532960448,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 22480
    },
    {
      "epoch": 8.981622053535757,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 22481
    },
    {
      "epoch": 8.982021574111066,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 22482
    },
    {
      "epoch": 8.982421094686377,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 22483
    },
    {
      "epoch": 8.982820615261685,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 22484
    },
    {
      "epoch": 8.983220135836996,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 22485
    },
    {
      "epoch": 8.983619656412305,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 22486
    },
    {
      "epoch": 8.984019176987616,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 22487
    },
    {
      "epoch": 8.984418697562925,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 22488
    },
    {
      "epoch": 8.984818218138233,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 22489
    },
    {
      "epoch": 8.985217738713544,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 22490
    },
    {
      "epoch": 8.985617259288853,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 22491
    },
    {
      "epoch": 8.986016779864164,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 22492
    },
    {
      "epoch": 8.986416300439473,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 22493
    },
    {
      "epoch": 8.986815821014781,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 22494
    },
    {
      "epoch": 8.987215341590092,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 22495
    },
    {
      "epoch": 8.987614862165401,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 22496
    },
    {
      "epoch": 8.988014382740712,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 22497
    },
    {
      "epoch": 8.98841390331602,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 22498
    },
    {
      "epoch": 8.98881342389133,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 22499
    },
    {
      "epoch": 8.98921294446664,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 22500
    },
    {
      "epoch": 8.989612465041949,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 22501
    },
    {
      "epoch": 8.99001198561726,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 22502
    },
    {
      "epoch": 8.990411506192569,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 22503
    },
    {
      "epoch": 8.99081102676788,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 22504
    },
    {
      "epoch": 8.991210547343188,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 22505
    },
    {
      "epoch": 8.991610067918497,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 22506
    },
    {
      "epoch": 8.992009588493808,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 22507
    },
    {
      "epoch": 8.992409109069117,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 22508
    },
    {
      "epoch": 8.992808629644427,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 22509
    },
    {
      "epoch": 8.993208150219736,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 22510
    },
    {
      "epoch": 8.993607670795045,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 22511
    },
    {
      "epoch": 8.994007191370356,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 22512
    },
    {
      "epoch": 8.994406711945665,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 22513
    },
    {
      "epoch": 8.994806232520975,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 22514
    },
    {
      "epoch": 8.995205753096284,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 22515
    },
    {
      "epoch": 8.995605273671593,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 22516
    },
    {
      "epoch": 8.996004794246904,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 22517
    },
    {
      "epoch": 8.996404314822213,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 22518
    },
    {
      "epoch": 8.996803835397523,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 22519
    },
    {
      "epoch": 8.997203355972832,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 22520
    },
    {
      "epoch": 8.997602876548143,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 22521
    },
    {
      "epoch": 8.998002397123452,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 22522
    },
    {
      "epoch": 8.99840191769876,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 22523
    },
    {
      "epoch": 8.998801438274072,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 22524
    },
    {
      "epoch": 8.99920095884938,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 22525
    },
    {
      "epoch": 8.999600479424691,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 22526
    },
    {
      "epoch": 9.0,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 22527
    },
    {
      "epoch": 9.000399520575309,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 22528
    },
    {
      "epoch": 9.00079904115062,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 22529
    },
    {
      "epoch": 9.001198561725928,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 22530
    },
    {
      "epoch": 9.00159808230124,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 22531
    },
    {
      "epoch": 9.001997602876548,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 22532
    },
    {
      "epoch": 9.002397123451857,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 22533
    },
    {
      "epoch": 9.002796644027168,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 22534
    },
    {
      "epoch": 9.003196164602477,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 22535
    },
    {
      "epoch": 9.003595685177787,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 22536
    },
    {
      "epoch": 9.003995205753096,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 22537
    },
    {
      "epoch": 9.004394726328407,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 22538
    },
    {
      "epoch": 9.004794246903716,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 22539
    },
    {
      "epoch": 9.005193767479025,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 22540
    },
    {
      "epoch": 9.005593288054335,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 22541
    },
    {
      "epoch": 9.005992808629644,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 22542
    },
    {
      "epoch": 9.006392329204955,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 22543
    },
    {
      "epoch": 9.006791849780264,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 22544
    },
    {
      "epoch": 9.007191370355573,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 22545
    },
    {
      "epoch": 9.007590890930883,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 22546
    },
    {
      "epoch": 9.007990411506192,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 22547
    },
    {
      "epoch": 9.008389932081503,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 22548
    },
    {
      "epoch": 9.008789452656812,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 22549
    },
    {
      "epoch": 9.00918897323212,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 22550
    },
    {
      "epoch": 9.009588493807431,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 22551
    },
    {
      "epoch": 9.00998801438274,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 22552
    },
    {
      "epoch": 9.010387534958051,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 22553
    },
    {
      "epoch": 9.01078705553336,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 22554
    },
    {
      "epoch": 9.01118657610867,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 22555
    },
    {
      "epoch": 9.01158609668398,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 22556
    },
    {
      "epoch": 9.011985617259288,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 22557
    },
    {
      "epoch": 9.012385137834599,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 22558
    },
    {
      "epoch": 9.012784658409908,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 22559
    },
    {
      "epoch": 9.013184178985219,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 22560
    },
    {
      "epoch": 9.013583699560527,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 22561
    },
    {
      "epoch": 9.013983220135836,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 22562
    },
    {
      "epoch": 9.014382740711147,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 22563
    },
    {
      "epoch": 9.014782261286456,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 22564
    },
    {
      "epoch": 9.015181781861767,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 22565
    },
    {
      "epoch": 9.015581302437075,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 22566
    },
    {
      "epoch": 9.015980823012384,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 22567
    },
    {
      "epoch": 9.016380343587695,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 22568
    },
    {
      "epoch": 9.016779864163004,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 22569
    },
    {
      "epoch": 9.017179384738315,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 22570
    },
    {
      "epoch": 9.017578905313623,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 22571
    },
    {
      "epoch": 9.017978425888934,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 22572
    },
    {
      "epoch": 9.018377946464243,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 22573
    },
    {
      "epoch": 9.018777467039552,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 22574
    },
    {
      "epoch": 9.019176987614863,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 22575
    },
    {
      "epoch": 9.019576508190172,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 22576
    },
    {
      "epoch": 9.019976028765482,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 22577
    },
    {
      "epoch": 9.020375549340791,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 22578
    },
    {
      "epoch": 9.0207750699161,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 22579
    },
    {
      "epoch": 9.02117459049141,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 22580
    },
    {
      "epoch": 9.02157411106672,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 22581
    },
    {
      "epoch": 9.02197363164203,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 22582
    },
    {
      "epoch": 9.02237315221734,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 22583
    },
    {
      "epoch": 9.022772672792648,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 22584
    },
    {
      "epoch": 9.023172193367959,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 22585
    },
    {
      "epoch": 9.023571713943268,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 22586
    },
    {
      "epoch": 9.023971234518578,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 22587
    },
    {
      "epoch": 9.024370755093887,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 22588
    },
    {
      "epoch": 9.024770275669196,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 22589
    },
    {
      "epoch": 9.025169796244507,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 22590
    },
    {
      "epoch": 9.025569316819816,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 22591
    },
    {
      "epoch": 9.025968837395126,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 22592
    },
    {
      "epoch": 9.026368357970435,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 22593
    },
    {
      "epoch": 9.026767878545746,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 22594
    },
    {
      "epoch": 9.027167399121055,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 22595
    },
    {
      "epoch": 9.027566919696364,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 22596
    },
    {
      "epoch": 9.027966440271674,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 22597
    },
    {
      "epoch": 9.028365960846983,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 22598
    },
    {
      "epoch": 9.028765481422294,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 22599
    },
    {
      "epoch": 9.029165001997603,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 22600
    },
    {
      "epoch": 9.029564522572912,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 22601
    },
    {
      "epoch": 9.029964043148222,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 22602
    },
    {
      "epoch": 9.030363563723531,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 22603
    },
    {
      "epoch": 9.030763084298842,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 22604
    },
    {
      "epoch": 9.031162604874151,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 22605
    },
    {
      "epoch": 9.03156212544946,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 22606
    },
    {
      "epoch": 9.03196164602477,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 22607
    },
    {
      "epoch": 9.03236116660008,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 22608
    },
    {
      "epoch": 9.03276068717539,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0557,
      "step": 22609
    },
    {
      "epoch": 9.033160207750699,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 22610
    },
    {
      "epoch": 9.03355972832601,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 22611
    },
    {
      "epoch": 9.033959248901319,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 22612
    },
    {
      "epoch": 9.034358769476627,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 22613
    },
    {
      "epoch": 9.034758290051938,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 22614
    },
    {
      "epoch": 9.035157810627247,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 22615
    },
    {
      "epoch": 9.035557331202558,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 22616
    },
    {
      "epoch": 9.035956851777867,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 22617
    },
    {
      "epoch": 9.036356372353175,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 22618
    },
    {
      "epoch": 9.036755892928486,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 22619
    },
    {
      "epoch": 9.037155413503795,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 22620
    },
    {
      "epoch": 9.037554934079106,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 22621
    },
    {
      "epoch": 9.037954454654415,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 22622
    },
    {
      "epoch": 9.038353975229724,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 22623
    },
    {
      "epoch": 9.038753495805034,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 22624
    },
    {
      "epoch": 9.039153016380343,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0526,
      "step": 22625
    },
    {
      "epoch": 9.039552536955654,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 22626
    },
    {
      "epoch": 9.039952057530963,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 22627
    },
    {
      "epoch": 9.040351578106273,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 22628
    },
    {
      "epoch": 9.040751098681582,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 22629
    },
    {
      "epoch": 9.041150619256891,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 22630
    },
    {
      "epoch": 9.041550139832202,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 22631
    },
    {
      "epoch": 9.04194966040751,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 22632
    },
    {
      "epoch": 9.042349180982821,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 22633
    },
    {
      "epoch": 9.04274870155813,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 22634
    },
    {
      "epoch": 9.04314822213344,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 22635
    },
    {
      "epoch": 9.04354774270875,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 22636
    },
    {
      "epoch": 9.043947263284059,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 22637
    },
    {
      "epoch": 9.04434678385937,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 22638
    },
    {
      "epoch": 9.044746304434678,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 22639
    },
    {
      "epoch": 9.045145825009987,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 22640
    },
    {
      "epoch": 9.045545345585298,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 22641
    },
    {
      "epoch": 9.045944866160607,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 22642
    },
    {
      "epoch": 9.046344386735917,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 22643
    },
    {
      "epoch": 9.046743907311226,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 22644
    },
    {
      "epoch": 9.047143427886537,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 22645
    },
    {
      "epoch": 9.047542948461846,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 22646
    },
    {
      "epoch": 9.047942469037155,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 22647
    },
    {
      "epoch": 9.048341989612465,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 22648
    },
    {
      "epoch": 9.048741510187774,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 22649
    },
    {
      "epoch": 9.049141030763085,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 22650
    },
    {
      "epoch": 9.049540551338394,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 22651
    },
    {
      "epoch": 9.049940071913703,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 22652
    },
    {
      "epoch": 9.050339592489014,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 22653
    },
    {
      "epoch": 9.050739113064322,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 22654
    },
    {
      "epoch": 9.051138633639633,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 22655
    },
    {
      "epoch": 9.051538154214942,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 22656
    },
    {
      "epoch": 9.051937674790251,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 22657
    },
    {
      "epoch": 9.052337195365562,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 22658
    },
    {
      "epoch": 9.05273671594087,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 22659
    },
    {
      "epoch": 9.053136236516181,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 22660
    },
    {
      "epoch": 9.05353575709149,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 22661
    },
    {
      "epoch": 9.053935277666799,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 22662
    },
    {
      "epoch": 9.05433479824211,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 22663
    },
    {
      "epoch": 9.054734318817419,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 22664
    },
    {
      "epoch": 9.05513383939273,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 22665
    },
    {
      "epoch": 9.055533359968038,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 22666
    },
    {
      "epoch": 9.055932880543349,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 22667
    },
    {
      "epoch": 9.056332401118658,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 22668
    },
    {
      "epoch": 9.056731921693967,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 22669
    },
    {
      "epoch": 9.057131442269277,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 22670
    },
    {
      "epoch": 9.057530962844586,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 22671
    },
    {
      "epoch": 9.057930483419897,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 22672
    },
    {
      "epoch": 9.058330003995206,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 22673
    },
    {
      "epoch": 9.058729524570515,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 22674
    },
    {
      "epoch": 9.059129045145825,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 22675
    },
    {
      "epoch": 9.059528565721134,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 22676
    },
    {
      "epoch": 9.059928086296445,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 22677
    },
    {
      "epoch": 9.060327606871754,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 22678
    },
    {
      "epoch": 9.060727127447063,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 22679
    },
    {
      "epoch": 9.061126648022373,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 22680
    },
    {
      "epoch": 9.061526168597682,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 22681
    },
    {
      "epoch": 9.061925689172993,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 22682
    },
    {
      "epoch": 9.062325209748302,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 22683
    },
    {
      "epoch": 9.062724730323612,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 22684
    },
    {
      "epoch": 9.063124250898921,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 22685
    },
    {
      "epoch": 9.06352377147423,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 22686
    },
    {
      "epoch": 9.063923292049541,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 22687
    },
    {
      "epoch": 9.06432281262485,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 22688
    },
    {
      "epoch": 9.06472233320016,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 22689
    },
    {
      "epoch": 9.06512185377547,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 22690
    },
    {
      "epoch": 9.065521374350778,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 22691
    },
    {
      "epoch": 9.065920894926089,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 22692
    },
    {
      "epoch": 9.066320415501398,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 22693
    },
    {
      "epoch": 9.066719936076709,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 22694
    },
    {
      "epoch": 9.067119456652017,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 22695
    },
    {
      "epoch": 9.067518977227326,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 22696
    },
    {
      "epoch": 9.067918497802637,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 22697
    },
    {
      "epoch": 9.068318018377946,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 22698
    },
    {
      "epoch": 9.068717538953257,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 22699
    },
    {
      "epoch": 9.069117059528566,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 22700
    },
    {
      "epoch": 9.069516580103876,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 22701
    },
    {
      "epoch": 9.069916100679185,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 22702
    },
    {
      "epoch": 9.070315621254494,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 22703
    },
    {
      "epoch": 9.070715141829805,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 22704
    },
    {
      "epoch": 9.071114662405114,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 22705
    },
    {
      "epoch": 9.071514182980424,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 22706
    },
    {
      "epoch": 9.071913703555733,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 22707
    },
    {
      "epoch": 9.072313224131042,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 22708
    },
    {
      "epoch": 9.072712744706353,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 22709
    },
    {
      "epoch": 9.073112265281662,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 22710
    },
    {
      "epoch": 9.073511785856972,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 22711
    },
    {
      "epoch": 9.073911306432281,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 22712
    },
    {
      "epoch": 9.07431082700759,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 22713
    },
    {
      "epoch": 9.0747103475829,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 22714
    },
    {
      "epoch": 9.07510986815821,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 22715
    },
    {
      "epoch": 9.07550938873352,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 22716
    },
    {
      "epoch": 9.07590890930883,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 22717
    },
    {
      "epoch": 9.07630842988414,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 22718
    },
    {
      "epoch": 9.076707950459449,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 22719
    },
    {
      "epoch": 9.077107471034758,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 22720
    },
    {
      "epoch": 9.077506991610068,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 22721
    },
    {
      "epoch": 9.077906512185377,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 22722
    },
    {
      "epoch": 9.078306032760688,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 22723
    },
    {
      "epoch": 9.078705553335997,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 22724
    },
    {
      "epoch": 9.079105073911306,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 22725
    },
    {
      "epoch": 9.079504594486616,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 22726
    },
    {
      "epoch": 9.079904115061925,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 22727
    },
    {
      "epoch": 9.080303635637236,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 22728
    },
    {
      "epoch": 9.080703156212545,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 22729
    },
    {
      "epoch": 9.081102676787854,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 22730
    },
    {
      "epoch": 9.081502197363164,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 22731
    },
    {
      "epoch": 9.081901717938473,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 22732
    },
    {
      "epoch": 9.082301238513784,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0544,
      "step": 22733
    },
    {
      "epoch": 9.082700759089093,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 22734
    },
    {
      "epoch": 9.083100279664404,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 22735
    },
    {
      "epoch": 9.083499800239712,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 22736
    },
    {
      "epoch": 9.083899320815021,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 22737
    },
    {
      "epoch": 9.084298841390332,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 22738
    },
    {
      "epoch": 9.084698361965641,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 22739
    },
    {
      "epoch": 9.085097882540952,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 22740
    },
    {
      "epoch": 9.08549740311626,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 22741
    },
    {
      "epoch": 9.08589692369157,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 22742
    },
    {
      "epoch": 9.08629644426688,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 22743
    },
    {
      "epoch": 9.086695964842189,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 22744
    },
    {
      "epoch": 9.0870954854175,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 22745
    },
    {
      "epoch": 9.087495005992809,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 22746
    },
    {
      "epoch": 9.087894526568117,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 22747
    },
    {
      "epoch": 9.088294047143428,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 22748
    },
    {
      "epoch": 9.088693567718737,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 22749
    },
    {
      "epoch": 9.089093088294048,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 22750
    },
    {
      "epoch": 9.089492608869357,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 22751
    },
    {
      "epoch": 9.089892129444667,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 22752
    },
    {
      "epoch": 9.090291650019976,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 22753
    },
    {
      "epoch": 9.090691170595285,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 22754
    },
    {
      "epoch": 9.091090691170596,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 22755
    },
    {
      "epoch": 9.091490211745905,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 22756
    },
    {
      "epoch": 9.091889732321215,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 22757
    },
    {
      "epoch": 9.092289252896524,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 22758
    },
    {
      "epoch": 9.092688773471833,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 22759
    },
    {
      "epoch": 9.093088294047144,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 22760
    },
    {
      "epoch": 9.093487814622453,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 22761
    },
    {
      "epoch": 9.093887335197763,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 22762
    },
    {
      "epoch": 9.094286855773072,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 22763
    },
    {
      "epoch": 9.094686376348381,
      "grad_norm": 0.296875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 22764
    },
    {
      "epoch": 9.095085896923692,
      "grad_norm": 0.734375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 22765
    },
    {
      "epoch": 9.095485417499,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 22766
    },
    {
      "epoch": 9.095884938074311,
      "grad_norm": 0.78125,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 22767
    },
    {
      "epoch": 9.09628445864962,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 22768
    },
    {
      "epoch": 9.09668397922493,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 22769
    },
    {
      "epoch": 9.09708349980024,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 22770
    },
    {
      "epoch": 9.097483020375549,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 22771
    },
    {
      "epoch": 9.09788254095086,
      "grad_norm": 0.46875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 22772
    },
    {
      "epoch": 9.098282061526168,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 22773
    },
    {
      "epoch": 9.098681582101479,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 22774
    },
    {
      "epoch": 9.099081102676788,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 22775
    },
    {
      "epoch": 9.099480623252097,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 22776
    },
    {
      "epoch": 9.099880143827408,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 22777
    },
    {
      "epoch": 9.100279664402716,
      "grad_norm": 0.40625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 22778
    },
    {
      "epoch": 9.100679184978027,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 22779
    },
    {
      "epoch": 9.101078705553336,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 22780
    },
    {
      "epoch": 9.101478226128645,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 22781
    },
    {
      "epoch": 9.101877746703956,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005,
      "loss": 1.0568,
      "step": 22782
    },
    {
      "epoch": 9.102277267279264,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 22783
    },
    {
      "epoch": 9.102676787854575,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 22784
    },
    {
      "epoch": 9.103076308429884,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 22785
    },
    {
      "epoch": 9.103475829005193,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 22786
    },
    {
      "epoch": 9.103875349580504,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 22787
    },
    {
      "epoch": 9.104274870155812,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 22788
    },
    {
      "epoch": 9.104674390731123,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 22789
    },
    {
      "epoch": 9.105073911306432,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 22790
    },
    {
      "epoch": 9.105473431881743,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 22791
    },
    {
      "epoch": 9.105872952457052,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 22792
    },
    {
      "epoch": 9.10627247303236,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 22793
    },
    {
      "epoch": 9.106671993607671,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 22794
    },
    {
      "epoch": 9.10707151418298,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 22795
    },
    {
      "epoch": 9.10747103475829,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 22796
    },
    {
      "epoch": 9.1078705553336,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 22797
    },
    {
      "epoch": 9.108270075908909,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 22798
    },
    {
      "epoch": 9.10866959648422,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 22799
    },
    {
      "epoch": 9.109069117059528,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 22800
    },
    {
      "epoch": 9.109468637634839,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 22801
    },
    {
      "epoch": 9.109868158210148,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 22802
    },
    {
      "epoch": 9.110267678785457,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 22803
    },
    {
      "epoch": 9.110667199360767,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 22804
    },
    {
      "epoch": 9.111066719936076,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 22805
    },
    {
      "epoch": 9.111466240511387,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 22806
    },
    {
      "epoch": 9.111865761086696,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 22807
    },
    {
      "epoch": 9.112265281662006,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 22808
    },
    {
      "epoch": 9.112664802237315,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 22809
    },
    {
      "epoch": 9.113064322812624,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 22810
    },
    {
      "epoch": 9.113463843387935,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 22811
    },
    {
      "epoch": 9.113863363963244,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 22812
    },
    {
      "epoch": 9.114262884538554,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 22813
    },
    {
      "epoch": 9.114662405113863,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 22814
    },
    {
      "epoch": 9.115061925689172,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 22815
    },
    {
      "epoch": 9.115461446264483,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 22816
    },
    {
      "epoch": 9.115860966839792,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 22817
    },
    {
      "epoch": 9.116260487415103,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 22818
    },
    {
      "epoch": 9.116660007990411,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 22819
    },
    {
      "epoch": 9.11705952856572,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 22820
    },
    {
      "epoch": 9.117459049141031,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 22821
    },
    {
      "epoch": 9.11785856971634,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 22822
    },
    {
      "epoch": 9.11825809029165,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 22823
    },
    {
      "epoch": 9.11865761086696,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 22824
    },
    {
      "epoch": 9.11905713144227,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 22825
    },
    {
      "epoch": 9.119456652017579,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 22826
    },
    {
      "epoch": 9.119856172592888,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 22827
    },
    {
      "epoch": 9.120255693168199,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 22828
    },
    {
      "epoch": 9.120655213743508,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 22829
    },
    {
      "epoch": 9.121054734318818,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 22830
    },
    {
      "epoch": 9.121454254894127,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 22831
    },
    {
      "epoch": 9.121853775469436,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 22832
    },
    {
      "epoch": 9.122253296044747,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 22833
    },
    {
      "epoch": 9.122652816620056,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 22834
    },
    {
      "epoch": 9.123052337195366,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 22835
    },
    {
      "epoch": 9.123451857770675,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 22836
    },
    {
      "epoch": 9.123851378345984,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 22837
    },
    {
      "epoch": 9.124250898921295,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 22838
    },
    {
      "epoch": 9.124650419496604,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 22839
    },
    {
      "epoch": 9.125049940071914,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 22840
    },
    {
      "epoch": 9.125449460647223,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 22841
    },
    {
      "epoch": 9.125848981222532,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 22842
    },
    {
      "epoch": 9.126248501797843,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 22843
    },
    {
      "epoch": 9.126648022373152,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 22844
    },
    {
      "epoch": 9.127047542948462,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 22845
    },
    {
      "epoch": 9.127447063523771,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 22846
    },
    {
      "epoch": 9.127846584099082,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 22847
    },
    {
      "epoch": 9.12824610467439,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 22848
    },
    {
      "epoch": 9.1286456252497,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 22849
    },
    {
      "epoch": 9.12904514582501,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 22850
    },
    {
      "epoch": 9.12944466640032,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 22851
    },
    {
      "epoch": 9.12984418697563,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 22852
    },
    {
      "epoch": 9.130243707550939,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 22853
    },
    {
      "epoch": 9.130643228126248,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 22854
    },
    {
      "epoch": 9.131042748701558,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 22855
    },
    {
      "epoch": 9.131442269276867,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 22856
    },
    {
      "epoch": 9.131841789852178,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 22857
    },
    {
      "epoch": 9.132241310427487,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 22858
    },
    {
      "epoch": 9.132640831002796,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 22859
    },
    {
      "epoch": 9.133040351578106,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 22860
    },
    {
      "epoch": 9.133439872153415,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 22861
    },
    {
      "epoch": 9.133839392728726,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 22862
    },
    {
      "epoch": 9.134238913304035,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 22863
    },
    {
      "epoch": 9.134638433879346,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 22864
    },
    {
      "epoch": 9.135037954454654,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 22865
    },
    {
      "epoch": 9.135437475029963,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 22866
    },
    {
      "epoch": 9.135836995605274,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 22867
    },
    {
      "epoch": 9.136236516180583,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 22868
    },
    {
      "epoch": 9.136636036755894,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 22869
    },
    {
      "epoch": 9.137035557331203,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 22870
    },
    {
      "epoch": 9.137435077906511,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 22871
    },
    {
      "epoch": 9.137834598481822,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 22872
    },
    {
      "epoch": 9.138234119057131,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 22873
    },
    {
      "epoch": 9.138633639632442,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 22874
    },
    {
      "epoch": 9.13903316020775,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 22875
    },
    {
      "epoch": 9.13943268078306,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 22876
    },
    {
      "epoch": 9.13983220135837,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 22877
    },
    {
      "epoch": 9.140231721933679,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 22878
    },
    {
      "epoch": 9.14063124250899,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 22879
    },
    {
      "epoch": 9.141030763084299,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 22880
    },
    {
      "epoch": 9.14143028365961,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 22881
    },
    {
      "epoch": 9.141829804234918,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 22882
    },
    {
      "epoch": 9.142229324810227,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 22883
    },
    {
      "epoch": 9.142628845385538,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 22884
    },
    {
      "epoch": 9.143028365960847,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 22885
    },
    {
      "epoch": 9.143427886536157,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 22886
    },
    {
      "epoch": 9.143827407111466,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 22887
    },
    {
      "epoch": 9.144226927686775,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 22888
    },
    {
      "epoch": 9.144626448262086,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 22889
    },
    {
      "epoch": 9.145025968837395,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 22890
    },
    {
      "epoch": 9.145425489412705,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 22891
    },
    {
      "epoch": 9.145825009988014,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 22892
    },
    {
      "epoch": 9.146224530563323,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 22893
    },
    {
      "epoch": 9.146624051138634,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 22894
    },
    {
      "epoch": 9.147023571713943,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 22895
    },
    {
      "epoch": 9.147423092289253,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 22896
    },
    {
      "epoch": 9.147822612864562,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 22897
    },
    {
      "epoch": 9.148222133439873,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 22898
    },
    {
      "epoch": 9.148621654015182,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 22899
    },
    {
      "epoch": 9.14902117459049,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 22900
    },
    {
      "epoch": 9.149420695165801,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 22901
    },
    {
      "epoch": 9.14982021574111,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 22902
    },
    {
      "epoch": 9.150219736316421,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 22903
    },
    {
      "epoch": 9.15061925689173,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 22904
    },
    {
      "epoch": 9.151018777467039,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 22905
    },
    {
      "epoch": 9.15141829804235,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 22906
    },
    {
      "epoch": 9.151817818617658,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 22907
    },
    {
      "epoch": 9.152217339192969,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 22908
    },
    {
      "epoch": 9.152616859768278,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 22909
    },
    {
      "epoch": 9.153016380343587,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 22910
    },
    {
      "epoch": 9.153415900918898,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 22911
    },
    {
      "epoch": 9.153815421494206,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 22912
    },
    {
      "epoch": 9.154214942069517,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 22913
    },
    {
      "epoch": 9.154614462644826,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 22914
    },
    {
      "epoch": 9.155013983220137,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 22915
    },
    {
      "epoch": 9.155413503795446,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 22916
    },
    {
      "epoch": 9.155813024370755,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 22917
    },
    {
      "epoch": 9.156212544946065,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 22918
    },
    {
      "epoch": 9.156612065521374,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 22919
    },
    {
      "epoch": 9.157011586096685,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 22920
    },
    {
      "epoch": 9.157411106671994,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 22921
    },
    {
      "epoch": 9.157810627247303,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 22922
    },
    {
      "epoch": 9.158210147822613,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 22923
    },
    {
      "epoch": 9.158609668397922,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 22924
    },
    {
      "epoch": 9.159009188973233,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 22925
    },
    {
      "epoch": 9.159408709548542,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 22926
    },
    {
      "epoch": 9.15980823012385,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 22927
    },
    {
      "epoch": 9.160207750699161,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 22928
    },
    {
      "epoch": 9.16060727127447,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 22929
    },
    {
      "epoch": 9.16100679184978,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 22930
    },
    {
      "epoch": 9.16140631242509,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 22931
    },
    {
      "epoch": 9.1618058330004,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 22932
    },
    {
      "epoch": 9.16220535357571,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 22933
    },
    {
      "epoch": 9.162604874151018,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 22934
    },
    {
      "epoch": 9.163004394726329,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 22935
    },
    {
      "epoch": 9.163403915301638,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 22936
    },
    {
      "epoch": 9.163803435876948,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 22937
    },
    {
      "epoch": 9.164202956452257,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 22938
    },
    {
      "epoch": 9.164602477027566,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 22939
    },
    {
      "epoch": 9.165001997602877,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 22940
    },
    {
      "epoch": 9.165401518178186,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 22941
    },
    {
      "epoch": 9.165801038753496,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 22942
    },
    {
      "epoch": 9.166200559328805,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 22943
    },
    {
      "epoch": 9.166600079904114,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 22944
    },
    {
      "epoch": 9.166999600479425,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 22945
    },
    {
      "epoch": 9.167399121054734,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 22946
    },
    {
      "epoch": 9.167798641630045,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 22947
    },
    {
      "epoch": 9.168198162205353,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 22948
    },
    {
      "epoch": 9.168597682780662,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 22949
    },
    {
      "epoch": 9.168997203355973,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 22950
    },
    {
      "epoch": 9.169396723931282,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 22951
    },
    {
      "epoch": 9.169796244506593,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 22952
    },
    {
      "epoch": 9.170195765081901,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 22953
    },
    {
      "epoch": 9.170595285657212,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 22954
    },
    {
      "epoch": 9.170994806232521,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 22955
    },
    {
      "epoch": 9.17139432680783,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 22956
    },
    {
      "epoch": 9.17179384738314,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 22957
    },
    {
      "epoch": 9.17219336795845,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 22958
    },
    {
      "epoch": 9.17259288853376,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 22959
    },
    {
      "epoch": 9.172992409109069,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 22960
    },
    {
      "epoch": 9.173391929684378,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 22961
    },
    {
      "epoch": 9.173791450259689,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 22962
    },
    {
      "epoch": 9.174190970834998,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 22963
    },
    {
      "epoch": 9.174590491410308,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 22964
    },
    {
      "epoch": 9.174990011985617,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 22965
    },
    {
      "epoch": 9.175389532560926,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 22966
    },
    {
      "epoch": 9.175789053136237,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 22967
    },
    {
      "epoch": 9.176188573711546,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 22968
    },
    {
      "epoch": 9.176588094286856,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 22969
    },
    {
      "epoch": 9.176987614862165,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 22970
    },
    {
      "epoch": 9.177387135437476,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 22971
    },
    {
      "epoch": 9.177786656012785,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 22972
    },
    {
      "epoch": 9.178186176588094,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 22973
    },
    {
      "epoch": 9.178585697163404,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 22974
    },
    {
      "epoch": 9.178985217738713,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 22975
    },
    {
      "epoch": 9.179384738314024,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 22976
    },
    {
      "epoch": 9.179784258889333,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 22977
    },
    {
      "epoch": 9.180183779464642,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 22978
    },
    {
      "epoch": 9.180583300039952,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 22979
    },
    {
      "epoch": 9.180982820615261,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 22980
    },
    {
      "epoch": 9.181382341190572,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 22981
    },
    {
      "epoch": 9.18178186176588,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 22982
    },
    {
      "epoch": 9.18218138234119,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 22983
    },
    {
      "epoch": 9.1825809029165,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 22984
    },
    {
      "epoch": 9.18298042349181,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 22985
    },
    {
      "epoch": 9.18337994406712,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 22986
    },
    {
      "epoch": 9.183779464642429,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 22987
    },
    {
      "epoch": 9.18417898521774,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 22988
    },
    {
      "epoch": 9.184578505793048,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 22989
    },
    {
      "epoch": 9.184978026368357,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 22990
    },
    {
      "epoch": 9.185377546943668,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 22991
    },
    {
      "epoch": 9.185777067518977,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 22992
    },
    {
      "epoch": 9.186176588094288,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 22993
    },
    {
      "epoch": 9.186576108669597,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 22994
    },
    {
      "epoch": 9.186975629244905,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 22995
    },
    {
      "epoch": 9.187375149820216,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 22996
    },
    {
      "epoch": 9.187774670395525,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 22997
    },
    {
      "epoch": 9.188174190970836,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 22998
    },
    {
      "epoch": 9.188573711546145,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 22999
    },
    {
      "epoch": 9.188973232121453,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 23000
    },
    {
      "epoch": 9.189372752696764,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 23001
    },
    {
      "epoch": 9.189772273272073,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 23002
    },
    {
      "epoch": 9.190171793847384,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 23003
    },
    {
      "epoch": 9.190571314422693,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 23004
    },
    {
      "epoch": 9.190970834998003,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 23005
    },
    {
      "epoch": 9.191370355573312,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 23006
    },
    {
      "epoch": 9.191769876148621,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 23007
    },
    {
      "epoch": 9.192169396723932,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 23008
    },
    {
      "epoch": 9.19256891729924,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 23009
    },
    {
      "epoch": 9.192968437874551,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 23010
    },
    {
      "epoch": 9.19336795844986,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 23011
    },
    {
      "epoch": 9.193767479025169,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 23012
    },
    {
      "epoch": 9.19416699960048,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 23013
    },
    {
      "epoch": 9.194566520175789,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 23014
    },
    {
      "epoch": 9.1949660407511,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 23015
    },
    {
      "epoch": 9.195365561326408,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 23016
    },
    {
      "epoch": 9.195765081901717,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 23017
    },
    {
      "epoch": 9.196164602477028,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 23018
    },
    {
      "epoch": 9.196564123052337,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 23019
    },
    {
      "epoch": 9.196963643627647,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 23020
    },
    {
      "epoch": 9.197363164202956,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 23021
    },
    {
      "epoch": 9.197762684778265,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 23022
    },
    {
      "epoch": 9.198162205353576,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 23023
    },
    {
      "epoch": 9.198561725928885,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 23024
    },
    {
      "epoch": 9.198961246504195,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 23025
    },
    {
      "epoch": 9.199360767079504,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 23026
    },
    {
      "epoch": 9.199760287654815,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 23027
    },
    {
      "epoch": 9.200159808230124,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 23028
    },
    {
      "epoch": 9.200559328805433,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0134,
      "step": 23029
    },
    {
      "epoch": 9.200958849380743,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 23030
    },
    {
      "epoch": 9.201358369956052,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 23031
    },
    {
      "epoch": 9.201757890531363,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 23032
    },
    {
      "epoch": 9.202157411106672,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 23033
    },
    {
      "epoch": 9.20255693168198,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 23034
    },
    {
      "epoch": 9.202956452257292,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 23035
    },
    {
      "epoch": 9.2033559728326,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 23036
    },
    {
      "epoch": 9.203755493407911,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 23037
    },
    {
      "epoch": 9.20415501398322,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 23038
    },
    {
      "epoch": 9.204554534558529,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 23039
    },
    {
      "epoch": 9.20495405513384,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 23040
    },
    {
      "epoch": 9.205353575709148,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 23041
    },
    {
      "epoch": 9.20575309628446,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 23042
    },
    {
      "epoch": 9.206152616859768,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 23043
    },
    {
      "epoch": 9.206552137435079,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 23044
    },
    {
      "epoch": 9.206951658010388,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 23045
    },
    {
      "epoch": 9.207351178585697,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 23046
    },
    {
      "epoch": 9.207750699161007,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 23047
    },
    {
      "epoch": 9.208150219736316,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 23048
    },
    {
      "epoch": 9.208549740311627,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 23049
    },
    {
      "epoch": 9.208949260886936,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 23050
    },
    {
      "epoch": 9.209348781462245,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 23051
    },
    {
      "epoch": 9.209748302037555,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 23052
    },
    {
      "epoch": 9.210147822612864,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 23053
    },
    {
      "epoch": 9.210547343188175,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 23054
    },
    {
      "epoch": 9.210946863763484,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 23055
    },
    {
      "epoch": 9.211346384338793,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 23056
    },
    {
      "epoch": 9.211745904914103,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 23057
    },
    {
      "epoch": 9.212145425489412,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 23058
    },
    {
      "epoch": 9.212544946064723,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 23059
    },
    {
      "epoch": 9.212944466640032,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 23060
    },
    {
      "epoch": 9.213343987215342,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 23061
    },
    {
      "epoch": 9.213743507790651,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 23062
    },
    {
      "epoch": 9.21414302836596,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 23063
    },
    {
      "epoch": 9.214542548941271,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 23064
    },
    {
      "epoch": 9.21494206951658,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 23065
    },
    {
      "epoch": 9.21534159009189,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 23066
    },
    {
      "epoch": 9.2157411106672,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 23067
    },
    {
      "epoch": 9.216140631242508,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 23068
    },
    {
      "epoch": 9.216540151817819,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 23069
    },
    {
      "epoch": 9.216939672393128,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 23070
    },
    {
      "epoch": 9.217339192968439,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 23071
    },
    {
      "epoch": 9.217738713543747,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 23072
    },
    {
      "epoch": 9.218138234119056,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 23073
    },
    {
      "epoch": 9.218537754694367,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 23074
    },
    {
      "epoch": 9.218937275269676,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 23075
    },
    {
      "epoch": 9.219336795844987,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 23076
    },
    {
      "epoch": 9.219736316420295,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 23077
    },
    {
      "epoch": 9.220135836995606,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 23078
    },
    {
      "epoch": 9.220535357570915,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 23079
    },
    {
      "epoch": 9.220934878146224,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 23080
    },
    {
      "epoch": 9.221334398721535,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 23081
    },
    {
      "epoch": 9.221733919296843,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 23082
    },
    {
      "epoch": 9.222133439872154,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 23083
    },
    {
      "epoch": 9.222532960447463,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 23084
    },
    {
      "epoch": 9.222932481022772,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 23085
    },
    {
      "epoch": 9.223332001598083,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 23086
    },
    {
      "epoch": 9.223731522173392,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 23087
    },
    {
      "epoch": 9.224131042748702,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 23088
    },
    {
      "epoch": 9.224530563324011,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 23089
    },
    {
      "epoch": 9.22493008389932,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 23090
    },
    {
      "epoch": 9.22532960447463,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 23091
    },
    {
      "epoch": 9.22572912504994,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 23092
    },
    {
      "epoch": 9.22612864562525,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 23093
    },
    {
      "epoch": 9.22652816620056,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 23094
    },
    {
      "epoch": 9.22692768677587,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 23095
    },
    {
      "epoch": 9.227327207351179,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 23096
    },
    {
      "epoch": 9.227726727926488,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 23097
    },
    {
      "epoch": 9.228126248501798,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 23098
    },
    {
      "epoch": 9.228525769077107,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 23099
    },
    {
      "epoch": 9.228925289652418,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 23100
    },
    {
      "epoch": 9.229324810227727,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 23101
    },
    {
      "epoch": 9.229724330803036,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 23102
    },
    {
      "epoch": 9.230123851378346,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 23103
    },
    {
      "epoch": 9.230523371953655,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 23104
    },
    {
      "epoch": 9.230922892528966,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 23105
    },
    {
      "epoch": 9.231322413104275,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 23106
    },
    {
      "epoch": 9.231721933679584,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 23107
    },
    {
      "epoch": 9.232121454254894,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 23108
    },
    {
      "epoch": 9.232520974830203,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 23109
    },
    {
      "epoch": 9.232920495405514,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 23110
    },
    {
      "epoch": 9.233320015980823,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 23111
    },
    {
      "epoch": 9.233719536556134,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 23112
    },
    {
      "epoch": 9.234119057131442,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 23113
    },
    {
      "epoch": 9.234518577706751,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 23114
    },
    {
      "epoch": 9.234918098282062,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 23115
    },
    {
      "epoch": 9.235317618857371,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 23116
    },
    {
      "epoch": 9.235717139432682,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 23117
    },
    {
      "epoch": 9.23611666000799,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 23118
    },
    {
      "epoch": 9.2365161805833,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 23119
    },
    {
      "epoch": 9.23691570115861,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 23120
    },
    {
      "epoch": 9.237315221733919,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 23121
    },
    {
      "epoch": 9.23771474230923,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 23122
    },
    {
      "epoch": 9.238114262884539,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 23123
    },
    {
      "epoch": 9.238513783459847,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 23124
    },
    {
      "epoch": 9.238913304035158,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 23125
    },
    {
      "epoch": 9.239312824610467,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 23126
    },
    {
      "epoch": 9.239712345185778,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 23127
    },
    {
      "epoch": 9.240111865761087,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 23128
    },
    {
      "epoch": 9.240511386336395,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 23129
    },
    {
      "epoch": 9.240910906911706,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 23130
    },
    {
      "epoch": 9.241310427487015,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 23131
    },
    {
      "epoch": 9.241709948062326,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 23132
    },
    {
      "epoch": 9.242109468637635,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 23133
    },
    {
      "epoch": 9.242508989212945,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0069,
      "step": 23134
    },
    {
      "epoch": 9.242908509788254,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 23135
    },
    {
      "epoch": 9.243308030363563,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 23136
    },
    {
      "epoch": 9.243707550938874,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 23137
    },
    {
      "epoch": 9.244107071514183,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 23138
    },
    {
      "epoch": 9.244506592089493,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 23139
    },
    {
      "epoch": 9.244906112664802,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 23140
    },
    {
      "epoch": 9.245305633240111,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 23141
    },
    {
      "epoch": 9.245705153815422,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 23142
    },
    {
      "epoch": 9.24610467439073,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 23143
    },
    {
      "epoch": 9.246504194966041,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 23144
    },
    {
      "epoch": 9.24690371554135,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 23145
    },
    {
      "epoch": 9.24730323611666,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 23146
    },
    {
      "epoch": 9.24770275669197,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 23147
    },
    {
      "epoch": 9.248102277267279,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 23148
    },
    {
      "epoch": 9.24850179784259,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 23149
    },
    {
      "epoch": 9.248901318417898,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 23150
    },
    {
      "epoch": 9.249300838993209,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 23151
    },
    {
      "epoch": 9.249700359568518,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 23152
    },
    {
      "epoch": 9.250099880143827,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 23153
    },
    {
      "epoch": 9.250499400719137,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 23154
    },
    {
      "epoch": 9.250898921294446,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 23155
    },
    {
      "epoch": 9.251298441869757,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 23156
    },
    {
      "epoch": 9.251697962445066,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 23157
    },
    {
      "epoch": 9.252097483020375,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 23158
    },
    {
      "epoch": 9.252497003595685,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 23159
    },
    {
      "epoch": 9.252896524170994,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 23160
    },
    {
      "epoch": 9.253296044746305,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 23161
    },
    {
      "epoch": 9.253695565321614,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 23162
    },
    {
      "epoch": 9.254095085896923,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 23163
    },
    {
      "epoch": 9.254494606472234,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 23164
    },
    {
      "epoch": 9.254894127047542,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 23165
    },
    {
      "epoch": 9.255293647622853,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 23166
    },
    {
      "epoch": 9.255693168198162,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 23167
    },
    {
      "epoch": 9.256092688773473,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 23168
    },
    {
      "epoch": 9.256492209348782,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 23169
    },
    {
      "epoch": 9.25689172992409,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 23170
    },
    {
      "epoch": 9.257291250499401,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 23171
    },
    {
      "epoch": 9.25769077107471,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 23172
    },
    {
      "epoch": 9.25809029165002,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 23173
    },
    {
      "epoch": 9.25848981222533,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 23174
    },
    {
      "epoch": 9.258889332800639,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 23175
    },
    {
      "epoch": 9.25928885337595,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 23176
    },
    {
      "epoch": 9.259688373951258,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 23177
    },
    {
      "epoch": 9.260087894526569,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 23178
    },
    {
      "epoch": 9.260487415101878,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 23179
    },
    {
      "epoch": 9.260886935677187,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 23180
    },
    {
      "epoch": 9.261286456252497,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 23181
    },
    {
      "epoch": 9.261685976827806,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 23182
    },
    {
      "epoch": 9.262085497403117,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 23183
    },
    {
      "epoch": 9.262485017978426,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 23184
    },
    {
      "epoch": 9.262884538553735,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 23185
    },
    {
      "epoch": 9.263284059129045,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 23186
    },
    {
      "epoch": 9.263683579704354,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 23187
    },
    {
      "epoch": 9.264083100279665,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 23188
    },
    {
      "epoch": 9.264482620854974,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 23189
    },
    {
      "epoch": 9.264882141430284,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 23190
    },
    {
      "epoch": 9.265281662005593,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 23191
    },
    {
      "epoch": 9.265681182580902,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 23192
    },
    {
      "epoch": 9.266080703156213,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 23193
    },
    {
      "epoch": 9.266480223731522,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 23194
    },
    {
      "epoch": 9.266879744306832,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 23195
    },
    {
      "epoch": 9.267279264882141,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 23196
    },
    {
      "epoch": 9.26767878545745,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 23197
    },
    {
      "epoch": 9.268078306032761,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 23198
    },
    {
      "epoch": 9.26847782660807,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 23199
    },
    {
      "epoch": 9.26887734718338,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 23200
    },
    {
      "epoch": 9.26927686775869,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 23201
    },
    {
      "epoch": 9.269676388333998,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 23202
    },
    {
      "epoch": 9.270075908909309,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 23203
    },
    {
      "epoch": 9.270475429484618,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 23204
    },
    {
      "epoch": 9.270874950059929,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 23205
    },
    {
      "epoch": 9.271274470635237,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 23206
    },
    {
      "epoch": 9.271673991210548,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 23207
    },
    {
      "epoch": 9.272073511785857,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 23208
    },
    {
      "epoch": 9.272473032361166,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 23209
    },
    {
      "epoch": 9.272872552936477,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 23210
    },
    {
      "epoch": 9.273272073511786,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 23211
    },
    {
      "epoch": 9.273671594087096,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 23212
    },
    {
      "epoch": 9.274071114662405,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 23213
    },
    {
      "epoch": 9.274470635237714,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 23214
    },
    {
      "epoch": 9.274870155813025,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 23215
    },
    {
      "epoch": 9.275269676388334,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 23216
    },
    {
      "epoch": 9.275669196963644,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 23217
    },
    {
      "epoch": 9.276068717538953,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 23218
    },
    {
      "epoch": 9.276468238114262,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 23219
    },
    {
      "epoch": 9.276867758689573,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 23220
    },
    {
      "epoch": 9.277267279264882,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 23221
    },
    {
      "epoch": 9.277666799840192,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 23222
    },
    {
      "epoch": 9.278066320415501,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 23223
    },
    {
      "epoch": 9.278465840990812,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 23224
    },
    {
      "epoch": 9.27886536156612,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 23225
    },
    {
      "epoch": 9.27926488214143,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 23226
    },
    {
      "epoch": 9.27966440271674,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 23227
    },
    {
      "epoch": 9.28006392329205,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 23228
    },
    {
      "epoch": 9.28046344386736,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 23229
    },
    {
      "epoch": 9.280862964442669,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 23230
    },
    {
      "epoch": 9.281262485017978,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 23231
    },
    {
      "epoch": 9.281662005593288,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 23232
    },
    {
      "epoch": 9.282061526168597,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 23233
    },
    {
      "epoch": 9.282461046743908,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 23234
    },
    {
      "epoch": 9.282860567319217,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 23235
    },
    {
      "epoch": 9.283260087894526,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 23236
    },
    {
      "epoch": 9.283659608469836,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 23237
    },
    {
      "epoch": 9.284059129045145,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 23238
    },
    {
      "epoch": 9.284458649620456,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 23239
    },
    {
      "epoch": 9.284858170195765,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 23240
    },
    {
      "epoch": 9.285257690771076,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 23241
    },
    {
      "epoch": 9.285657211346384,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 23242
    },
    {
      "epoch": 9.286056731921693,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 23243
    },
    {
      "epoch": 9.286456252497004,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 23244
    },
    {
      "epoch": 9.286855773072313,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 23245
    },
    {
      "epoch": 9.287255293647624,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 23246
    },
    {
      "epoch": 9.287654814222932,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 23247
    },
    {
      "epoch": 9.288054334798241,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 23248
    },
    {
      "epoch": 9.288453855373552,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 23249
    },
    {
      "epoch": 9.288853375948861,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 23250
    },
    {
      "epoch": 9.289252896524172,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 23251
    },
    {
      "epoch": 9.28965241709948,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 23252
    },
    {
      "epoch": 9.29005193767479,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 23253
    },
    {
      "epoch": 9.2904514582501,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 23254
    },
    {
      "epoch": 9.290850978825409,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 23255
    },
    {
      "epoch": 9.29125049940072,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 23256
    },
    {
      "epoch": 9.291650019976029,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 23257
    },
    {
      "epoch": 9.29204954055134,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 23258
    },
    {
      "epoch": 9.292449061126648,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 23259
    },
    {
      "epoch": 9.292848581701957,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 23260
    },
    {
      "epoch": 9.293248102277268,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 23261
    },
    {
      "epoch": 9.293647622852577,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 23262
    },
    {
      "epoch": 9.294047143427887,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 23263
    },
    {
      "epoch": 9.294446664003196,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 23264
    },
    {
      "epoch": 9.294846184578505,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 23265
    },
    {
      "epoch": 9.295245705153816,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 23266
    },
    {
      "epoch": 9.295645225729125,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 23267
    },
    {
      "epoch": 9.296044746304435,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 23268
    },
    {
      "epoch": 9.296444266879744,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 23269
    },
    {
      "epoch": 9.296843787455053,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 23270
    },
    {
      "epoch": 9.297243308030364,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 23271
    },
    {
      "epoch": 9.297642828605673,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 23272
    },
    {
      "epoch": 9.298042349180983,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 23273
    },
    {
      "epoch": 9.298441869756292,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 23274
    },
    {
      "epoch": 9.298841390331603,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 23275
    },
    {
      "epoch": 9.299240910906912,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 23276
    },
    {
      "epoch": 9.29964043148222,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 23277
    },
    {
      "epoch": 9.300039952057531,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 23278
    },
    {
      "epoch": 9.30043947263284,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 23279
    },
    {
      "epoch": 9.300838993208151,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 23280
    },
    {
      "epoch": 9.30123851378346,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 23281
    },
    {
      "epoch": 9.301638034358769,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 23282
    },
    {
      "epoch": 9.30203755493408,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 23283
    },
    {
      "epoch": 9.302437075509388,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 23284
    },
    {
      "epoch": 9.302836596084699,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 23285
    },
    {
      "epoch": 9.303236116660008,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 23286
    },
    {
      "epoch": 9.303635637235317,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 23287
    },
    {
      "epoch": 9.304035157810628,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 23288
    },
    {
      "epoch": 9.304434678385936,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 23289
    },
    {
      "epoch": 9.304834198961247,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 23290
    },
    {
      "epoch": 9.305233719536556,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 23291
    },
    {
      "epoch": 9.305633240111867,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 23292
    },
    {
      "epoch": 9.306032760687176,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 23293
    },
    {
      "epoch": 9.306432281262484,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 23294
    },
    {
      "epoch": 9.306831801837795,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 23295
    },
    {
      "epoch": 9.307231322413104,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 23296
    },
    {
      "epoch": 9.307630842988415,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 23297
    },
    {
      "epoch": 9.308030363563724,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 23298
    },
    {
      "epoch": 9.308429884139032,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 23299
    },
    {
      "epoch": 9.308829404714343,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 23300
    },
    {
      "epoch": 9.309228925289652,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 23301
    },
    {
      "epoch": 9.309628445864963,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 23302
    },
    {
      "epoch": 9.310027966440272,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 23303
    },
    {
      "epoch": 9.31042748701558,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 23304
    },
    {
      "epoch": 9.310827007590891,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 23305
    },
    {
      "epoch": 9.3112265281662,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 23306
    },
    {
      "epoch": 9.31162604874151,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 23307
    },
    {
      "epoch": 9.31202556931682,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 23308
    },
    {
      "epoch": 9.31242508989213,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 23309
    },
    {
      "epoch": 9.31282461046744,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 23310
    },
    {
      "epoch": 9.313224131042748,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 23311
    },
    {
      "epoch": 9.313623651618059,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 23312
    },
    {
      "epoch": 9.314023172193368,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 23313
    },
    {
      "epoch": 9.314422692768678,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 23314
    },
    {
      "epoch": 9.314822213343987,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 23315
    },
    {
      "epoch": 9.315221733919296,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 23316
    },
    {
      "epoch": 9.315621254494607,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 23317
    },
    {
      "epoch": 9.316020775069916,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 23318
    },
    {
      "epoch": 9.316420295645226,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 23319
    },
    {
      "epoch": 9.316819816220535,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 23320
    },
    {
      "epoch": 9.317219336795844,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 23321
    },
    {
      "epoch": 9.317618857371155,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 23322
    },
    {
      "epoch": 9.318018377946464,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 23323
    },
    {
      "epoch": 9.318417898521774,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 23324
    },
    {
      "epoch": 9.318817419097083,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 23325
    },
    {
      "epoch": 9.319216939672392,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 23326
    },
    {
      "epoch": 9.319616460247703,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 23327
    },
    {
      "epoch": 9.320015980823012,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 23328
    },
    {
      "epoch": 9.320415501398323,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 23329
    },
    {
      "epoch": 9.320815021973631,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 23330
    },
    {
      "epoch": 9.321214542548942,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 23331
    },
    {
      "epoch": 9.321614063124251,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0552,
      "step": 23332
    },
    {
      "epoch": 9.32201358369956,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 23333
    },
    {
      "epoch": 9.32241310427487,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 23334
    },
    {
      "epoch": 9.32281262485018,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 23335
    },
    {
      "epoch": 9.32321214542549,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 23336
    },
    {
      "epoch": 9.323611666000799,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 23337
    },
    {
      "epoch": 9.324011186576108,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 23338
    },
    {
      "epoch": 9.324410707151419,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 23339
    },
    {
      "epoch": 9.324810227726728,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 23340
    },
    {
      "epoch": 9.325209748302038,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 23341
    },
    {
      "epoch": 9.325609268877347,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 23342
    },
    {
      "epoch": 9.326008789452656,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 23343
    },
    {
      "epoch": 9.326408310027967,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 23344
    },
    {
      "epoch": 9.326807830603276,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 23345
    },
    {
      "epoch": 9.327207351178586,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 23346
    },
    {
      "epoch": 9.327606871753895,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 23347
    },
    {
      "epoch": 9.328006392329206,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 23348
    },
    {
      "epoch": 9.328405912904515,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 23349
    },
    {
      "epoch": 9.328805433479824,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 23350
    },
    {
      "epoch": 9.329204954055134,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 23351
    },
    {
      "epoch": 9.329604474630443,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 23352
    },
    {
      "epoch": 9.330003995205754,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 23353
    },
    {
      "epoch": 9.330403515781063,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 23354
    },
    {
      "epoch": 9.330803036356372,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 23355
    },
    {
      "epoch": 9.331202556931682,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 23356
    },
    {
      "epoch": 9.331602077506991,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 23357
    },
    {
      "epoch": 9.332001598082302,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 23358
    },
    {
      "epoch": 9.33240111865761,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 23359
    },
    {
      "epoch": 9.33280063923292,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 23360
    },
    {
      "epoch": 9.33320015980823,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 23361
    },
    {
      "epoch": 9.33359968038354,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 23362
    },
    {
      "epoch": 9.33399920095885,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 23363
    },
    {
      "epoch": 9.334398721534159,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 23364
    },
    {
      "epoch": 9.334798242109468,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 23365
    },
    {
      "epoch": 9.335197762684778,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 23366
    },
    {
      "epoch": 9.335597283260087,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 23367
    },
    {
      "epoch": 9.335996803835398,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 23368
    },
    {
      "epoch": 9.336396324410707,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 23369
    },
    {
      "epoch": 9.336795844986018,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 23370
    },
    {
      "epoch": 9.337195365561326,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 23371
    },
    {
      "epoch": 9.337594886136635,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 23372
    },
    {
      "epoch": 9.337994406711946,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 23373
    },
    {
      "epoch": 9.338393927287255,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 23374
    },
    {
      "epoch": 9.338793447862566,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 23375
    },
    {
      "epoch": 9.339192968437874,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 23376
    },
    {
      "epoch": 9.339592489013183,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 23377
    },
    {
      "epoch": 9.339992009588494,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 23378
    },
    {
      "epoch": 9.340391530163803,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 23379
    },
    {
      "epoch": 9.340791050739114,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 23380
    },
    {
      "epoch": 9.341190571314423,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 23381
    },
    {
      "epoch": 9.341590091889731,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 23382
    },
    {
      "epoch": 9.341989612465042,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 23383
    },
    {
      "epoch": 9.342389133040351,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 23384
    },
    {
      "epoch": 9.342788653615662,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 23385
    },
    {
      "epoch": 9.34318817419097,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 23386
    },
    {
      "epoch": 9.343587694766281,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 23387
    },
    {
      "epoch": 9.34398721534159,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 23388
    },
    {
      "epoch": 9.344386735916899,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 23389
    },
    {
      "epoch": 9.34478625649221,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 23390
    },
    {
      "epoch": 9.345185777067519,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 23391
    },
    {
      "epoch": 9.34558529764283,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 23392
    },
    {
      "epoch": 9.345984818218138,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 23393
    },
    {
      "epoch": 9.346384338793447,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 23394
    },
    {
      "epoch": 9.346783859368758,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 23395
    },
    {
      "epoch": 9.347183379944067,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 23396
    },
    {
      "epoch": 9.347582900519377,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 23397
    },
    {
      "epoch": 9.347982421094686,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 23398
    },
    {
      "epoch": 9.348381941669995,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 23399
    },
    {
      "epoch": 9.348781462245306,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 23400
    },
    {
      "epoch": 9.349180982820615,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 23401
    },
    {
      "epoch": 9.349580503395925,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 23402
    },
    {
      "epoch": 9.349980023971234,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 23403
    },
    {
      "epoch": 9.350379544546545,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 23404
    },
    {
      "epoch": 9.350779065121854,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 23405
    },
    {
      "epoch": 9.351178585697163,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 23406
    },
    {
      "epoch": 9.351578106272473,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 23407
    },
    {
      "epoch": 9.351977626847782,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 23408
    },
    {
      "epoch": 9.352377147423093,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 23409
    },
    {
      "epoch": 9.352776667998402,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 23410
    },
    {
      "epoch": 9.35317618857371,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 23411
    },
    {
      "epoch": 9.353575709149021,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 23412
    },
    {
      "epoch": 9.35397522972433,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 23413
    },
    {
      "epoch": 9.354374750299641,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 23414
    },
    {
      "epoch": 9.35477427087495,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 23415
    },
    {
      "epoch": 9.355173791450259,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 23416
    },
    {
      "epoch": 9.35557331202557,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 23417
    },
    {
      "epoch": 9.355972832600878,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 23418
    },
    {
      "epoch": 9.356372353176189,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 23419
    },
    {
      "epoch": 9.356771873751498,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 23420
    },
    {
      "epoch": 9.357171394326809,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 23421
    },
    {
      "epoch": 9.357570914902118,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 23422
    },
    {
      "epoch": 9.357970435477426,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 23423
    },
    {
      "epoch": 9.358369956052737,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 23424
    },
    {
      "epoch": 9.358769476628046,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 23425
    },
    {
      "epoch": 9.359168997203357,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 23426
    },
    {
      "epoch": 9.359568517778666,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 23427
    },
    {
      "epoch": 9.359968038353975,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 23428
    },
    {
      "epoch": 9.360367558929285,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 23429
    },
    {
      "epoch": 9.360767079504594,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 23430
    },
    {
      "epoch": 9.361166600079905,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 23431
    },
    {
      "epoch": 9.361566120655214,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 23432
    },
    {
      "epoch": 9.361965641230523,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 23433
    },
    {
      "epoch": 9.362365161805833,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 23434
    },
    {
      "epoch": 9.362764682381142,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 23435
    },
    {
      "epoch": 9.363164202956453,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 23436
    },
    {
      "epoch": 9.363563723531762,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 23437
    },
    {
      "epoch": 9.363963244107072,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 23438
    },
    {
      "epoch": 9.364362764682381,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 23439
    },
    {
      "epoch": 9.36476228525769,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 23440
    },
    {
      "epoch": 9.365161805833,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 23441
    },
    {
      "epoch": 9.36556132640831,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 23442
    },
    {
      "epoch": 9.36596084698362,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 23443
    },
    {
      "epoch": 9.36636036755893,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 23444
    },
    {
      "epoch": 9.366759888134238,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 23445
    },
    {
      "epoch": 9.367159408709549,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 23446
    },
    {
      "epoch": 9.367558929284858,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 23447
    },
    {
      "epoch": 9.367958449860168,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 23448
    },
    {
      "epoch": 9.368357970435477,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 23449
    },
    {
      "epoch": 9.368757491010786,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 23450
    },
    {
      "epoch": 9.369157011586097,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 23451
    },
    {
      "epoch": 9.369556532161406,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 23452
    },
    {
      "epoch": 9.369956052736717,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 23453
    },
    {
      "epoch": 9.370355573312025,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 23454
    },
    {
      "epoch": 9.370755093887336,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 23455
    },
    {
      "epoch": 9.371154614462645,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 23456
    },
    {
      "epoch": 9.371554135037954,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 23457
    },
    {
      "epoch": 9.371953655613265,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 23458
    },
    {
      "epoch": 9.372353176188573,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 23459
    },
    {
      "epoch": 9.372752696763884,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 23460
    },
    {
      "epoch": 9.373152217339193,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 23461
    },
    {
      "epoch": 9.373551737914502,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 23462
    },
    {
      "epoch": 9.373951258489813,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 23463
    },
    {
      "epoch": 9.374350779065121,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 23464
    },
    {
      "epoch": 9.374750299640432,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 23465
    },
    {
      "epoch": 9.375149820215741,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 23466
    },
    {
      "epoch": 9.37554934079105,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 23467
    },
    {
      "epoch": 9.37594886136636,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 23468
    },
    {
      "epoch": 9.37634838194167,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 23469
    },
    {
      "epoch": 9.37674790251698,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 23470
    },
    {
      "epoch": 9.377147423092289,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 23471
    },
    {
      "epoch": 9.3775469436676,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 23472
    },
    {
      "epoch": 9.377946464242909,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 23473
    },
    {
      "epoch": 9.378345984818218,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 23474
    },
    {
      "epoch": 9.378745505393528,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 23475
    },
    {
      "epoch": 9.379145025968837,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 23476
    },
    {
      "epoch": 9.379544546544148,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 23477
    },
    {
      "epoch": 9.379944067119457,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 23478
    },
    {
      "epoch": 9.380343587694766,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 23479
    },
    {
      "epoch": 9.380743108270076,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 23480
    },
    {
      "epoch": 9.381142628845385,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 23481
    },
    {
      "epoch": 9.381542149420696,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 23482
    },
    {
      "epoch": 9.381941669996005,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 23483
    },
    {
      "epoch": 9.382341190571314,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 23484
    },
    {
      "epoch": 9.382740711146624,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 23485
    },
    {
      "epoch": 9.383140231721933,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 23486
    },
    {
      "epoch": 9.383539752297244,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 23487
    },
    {
      "epoch": 9.383939272872553,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 23488
    },
    {
      "epoch": 9.384338793447863,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 23489
    },
    {
      "epoch": 9.384738314023172,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 23490
    },
    {
      "epoch": 9.385137834598481,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 23491
    },
    {
      "epoch": 9.385537355173792,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 23492
    },
    {
      "epoch": 9.3859368757491,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 23493
    },
    {
      "epoch": 9.386336396324412,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 23494
    },
    {
      "epoch": 9.38673591689972,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 23495
    },
    {
      "epoch": 9.38713543747503,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 23496
    },
    {
      "epoch": 9.38753495805034,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 23497
    },
    {
      "epoch": 9.387934478625649,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 23498
    },
    {
      "epoch": 9.38833399920096,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 23499
    },
    {
      "epoch": 9.388733519776268,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 23500
    },
    {
      "epoch": 9.389133040351577,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 23501
    },
    {
      "epoch": 9.389532560926888,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 23502
    },
    {
      "epoch": 9.389932081502197,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 23503
    },
    {
      "epoch": 9.390331602077508,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 23504
    },
    {
      "epoch": 9.390731122652817,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 23505
    },
    {
      "epoch": 9.391130643228125,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 23506
    },
    {
      "epoch": 9.391530163803436,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 23507
    },
    {
      "epoch": 9.391929684378745,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 23508
    },
    {
      "epoch": 9.392329204954056,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 23509
    },
    {
      "epoch": 9.392728725529365,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 23510
    },
    {
      "epoch": 9.393128246104675,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 23511
    },
    {
      "epoch": 9.393527766679984,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 23512
    },
    {
      "epoch": 9.393927287255293,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 23513
    },
    {
      "epoch": 9.394326807830604,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 23514
    },
    {
      "epoch": 9.394726328405913,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 23515
    },
    {
      "epoch": 9.395125848981223,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 23516
    },
    {
      "epoch": 9.395525369556532,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 23517
    },
    {
      "epoch": 9.395924890131841,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 23518
    },
    {
      "epoch": 9.396324410707152,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 23519
    },
    {
      "epoch": 9.39672393128246,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 23520
    },
    {
      "epoch": 9.397123451857771,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 23521
    },
    {
      "epoch": 9.39752297243308,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 23522
    },
    {
      "epoch": 9.397922493008389,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 23523
    },
    {
      "epoch": 9.3983220135837,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 23524
    },
    {
      "epoch": 9.398721534159009,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 23525
    },
    {
      "epoch": 9.39912105473432,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 23526
    },
    {
      "epoch": 9.399520575309628,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 23527
    },
    {
      "epoch": 9.399920095884939,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 23528
    },
    {
      "epoch": 9.400319616460248,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 23529
    },
    {
      "epoch": 9.400719137035557,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 23530
    },
    {
      "epoch": 9.401118657610867,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 23531
    },
    {
      "epoch": 9.401518178186176,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 23532
    },
    {
      "epoch": 9.401917698761487,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 23533
    },
    {
      "epoch": 9.402317219336796,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 23534
    },
    {
      "epoch": 9.402716739912105,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 23535
    },
    {
      "epoch": 9.403116260487415,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 23536
    },
    {
      "epoch": 9.403515781062724,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 23537
    },
    {
      "epoch": 9.403915301638035,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 23538
    },
    {
      "epoch": 9.404314822213344,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 23539
    },
    {
      "epoch": 9.404714342788653,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 23540
    },
    {
      "epoch": 9.405113863363963,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 23541
    },
    {
      "epoch": 9.405513383939272,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 23542
    },
    {
      "epoch": 9.405912904514583,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 23543
    },
    {
      "epoch": 9.406312425089892,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 23544
    },
    {
      "epoch": 9.4067119456652,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 23545
    },
    {
      "epoch": 9.407111466240512,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 23546
    },
    {
      "epoch": 9.40751098681582,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 23547
    },
    {
      "epoch": 9.407910507391131,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 23548
    },
    {
      "epoch": 9.40831002796644,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 23549
    },
    {
      "epoch": 9.40870954854175,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 23550
    },
    {
      "epoch": 9.40910906911706,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 23551
    },
    {
      "epoch": 9.409508589692368,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 23552
    },
    {
      "epoch": 9.40990811026768,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 23553
    },
    {
      "epoch": 9.410307630842988,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 23554
    },
    {
      "epoch": 9.410707151418299,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 23555
    },
    {
      "epoch": 9.411106671993608,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 23556
    },
    {
      "epoch": 9.411506192568917,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 23557
    },
    {
      "epoch": 9.411905713144227,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 23558
    },
    {
      "epoch": 9.412305233719536,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 23559
    },
    {
      "epoch": 9.412704754294847,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 23560
    },
    {
      "epoch": 9.413104274870156,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 23561
    },
    {
      "epoch": 9.413503795445465,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 23562
    },
    {
      "epoch": 9.413903316020775,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 23563
    },
    {
      "epoch": 9.414302836596084,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 23564
    },
    {
      "epoch": 9.414702357171395,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 23565
    },
    {
      "epoch": 9.415101877746704,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 23566
    },
    {
      "epoch": 9.415501398322014,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 23567
    },
    {
      "epoch": 9.415900918897323,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 23568
    },
    {
      "epoch": 9.416300439472632,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 23569
    },
    {
      "epoch": 9.416699960047943,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 23570
    },
    {
      "epoch": 9.417099480623252,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 23571
    },
    {
      "epoch": 9.417499001198562,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 23572
    },
    {
      "epoch": 9.417898521773871,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 23573
    },
    {
      "epoch": 9.41829804234918,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 23574
    },
    {
      "epoch": 9.418697562924491,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 23575
    },
    {
      "epoch": 9.4190970834998,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 23576
    },
    {
      "epoch": 9.41949660407511,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 23577
    },
    {
      "epoch": 9.41989612465042,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 23578
    },
    {
      "epoch": 9.420295645225728,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 23579
    },
    {
      "epoch": 9.420695165801039,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 23580
    },
    {
      "epoch": 9.421094686376348,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 23581
    },
    {
      "epoch": 9.421494206951659,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 23582
    },
    {
      "epoch": 9.421893727526967,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 23583
    },
    {
      "epoch": 9.422293248102278,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 23584
    },
    {
      "epoch": 9.422692768677587,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 23585
    },
    {
      "epoch": 9.423092289252896,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 23586
    },
    {
      "epoch": 9.423491809828207,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 23587
    },
    {
      "epoch": 9.423891330403515,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 23588
    },
    {
      "epoch": 9.424290850978826,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 23589
    },
    {
      "epoch": 9.424690371554135,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 23590
    },
    {
      "epoch": 9.425089892129444,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 23591
    },
    {
      "epoch": 9.425489412704755,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 23592
    },
    {
      "epoch": 9.425888933280063,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 23593
    },
    {
      "epoch": 9.426288453855374,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 23594
    },
    {
      "epoch": 9.426687974430683,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 23595
    },
    {
      "epoch": 9.427087495005992,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 23596
    },
    {
      "epoch": 9.427487015581303,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 23597
    },
    {
      "epoch": 9.427886536156612,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 23598
    },
    {
      "epoch": 9.428286056731922,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 23599
    },
    {
      "epoch": 9.428685577307231,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 23600
    },
    {
      "epoch": 9.429085097882542,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 23601
    },
    {
      "epoch": 9.42948461845785,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 23602
    },
    {
      "epoch": 9.42988413903316,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 23603
    },
    {
      "epoch": 9.43028365960847,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 23604
    },
    {
      "epoch": 9.43068318018378,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 23605
    },
    {
      "epoch": 9.43108270075909,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 23606
    },
    {
      "epoch": 9.431482221334399,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 23607
    },
    {
      "epoch": 9.431881741909708,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 23608
    },
    {
      "epoch": 9.432281262485018,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 23609
    },
    {
      "epoch": 9.432680783060327,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 23610
    },
    {
      "epoch": 9.433080303635638,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 23611
    },
    {
      "epoch": 9.433479824210947,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 23612
    },
    {
      "epoch": 9.433879344786256,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 23613
    },
    {
      "epoch": 9.434278865361566,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 23614
    },
    {
      "epoch": 9.434678385936875,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 23615
    },
    {
      "epoch": 9.435077906512186,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 23616
    },
    {
      "epoch": 9.435477427087495,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 23617
    },
    {
      "epoch": 9.435876947662805,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 23618
    },
    {
      "epoch": 9.436276468238114,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 23619
    },
    {
      "epoch": 9.436675988813423,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 23620
    },
    {
      "epoch": 9.437075509388734,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 23621
    },
    {
      "epoch": 9.437475029964043,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 23622
    },
    {
      "epoch": 9.437874550539354,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 23623
    },
    {
      "epoch": 9.438274071114662,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 23624
    },
    {
      "epoch": 9.438673591689971,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 23625
    },
    {
      "epoch": 9.439073112265282,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 23626
    },
    {
      "epoch": 9.439472632840591,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 23627
    },
    {
      "epoch": 9.439872153415902,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 23628
    },
    {
      "epoch": 9.44027167399121,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 23629
    },
    {
      "epoch": 9.44067119456652,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 23630
    },
    {
      "epoch": 9.44107071514183,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 23631
    },
    {
      "epoch": 9.441470235717139,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 23632
    },
    {
      "epoch": 9.44186975629245,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 23633
    },
    {
      "epoch": 9.442269276867759,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 23634
    },
    {
      "epoch": 9.44266879744307,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 23635
    },
    {
      "epoch": 9.443068318018378,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 23636
    },
    {
      "epoch": 9.443467838593687,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 23637
    },
    {
      "epoch": 9.443867359168998,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 23638
    },
    {
      "epoch": 9.444266879744307,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 23639
    },
    {
      "epoch": 9.444666400319617,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 23640
    },
    {
      "epoch": 9.445065920894926,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 23641
    },
    {
      "epoch": 9.445465441470235,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 23642
    },
    {
      "epoch": 9.445864962045546,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 23643
    },
    {
      "epoch": 9.446264482620855,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 23644
    },
    {
      "epoch": 9.446664003196165,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 23645
    },
    {
      "epoch": 9.447063523771474,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 23646
    },
    {
      "epoch": 9.447463044346783,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 23647
    },
    {
      "epoch": 9.447862564922094,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 23648
    },
    {
      "epoch": 9.448262085497403,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 23649
    },
    {
      "epoch": 9.448661606072713,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 23650
    },
    {
      "epoch": 9.449061126648022,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 23651
    },
    {
      "epoch": 9.449460647223333,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 23652
    },
    {
      "epoch": 9.449860167798642,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 23653
    },
    {
      "epoch": 9.45025968837395,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 23654
    },
    {
      "epoch": 9.450659208949261,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 23655
    },
    {
      "epoch": 9.45105872952457,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 23656
    },
    {
      "epoch": 9.451458250099881,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 23657
    },
    {
      "epoch": 9.45185777067519,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 23658
    },
    {
      "epoch": 9.452257291250499,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 23659
    },
    {
      "epoch": 9.45265681182581,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 23660
    },
    {
      "epoch": 9.453056332401118,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 23661
    },
    {
      "epoch": 9.453455852976429,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 23662
    },
    {
      "epoch": 9.453855373551738,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 23663
    },
    {
      "epoch": 9.454254894127047,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 23664
    },
    {
      "epoch": 9.454654414702357,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 23665
    },
    {
      "epoch": 9.455053935277666,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 23666
    },
    {
      "epoch": 9.455453455852977,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 23667
    },
    {
      "epoch": 9.455852976428286,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 23668
    },
    {
      "epoch": 9.456252497003595,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 23669
    },
    {
      "epoch": 9.456652017578906,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 23670
    },
    {
      "epoch": 9.457051538154214,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 23671
    },
    {
      "epoch": 9.457451058729525,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 23672
    },
    {
      "epoch": 9.457850579304834,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 23673
    },
    {
      "epoch": 9.458250099880145,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 23674
    },
    {
      "epoch": 9.458649620455454,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 23675
    },
    {
      "epoch": 9.459049141030762,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 23676
    },
    {
      "epoch": 9.459448661606073,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 23677
    },
    {
      "epoch": 9.459848182181382,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 23678
    },
    {
      "epoch": 9.460247702756693,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 23679
    },
    {
      "epoch": 9.460647223332002,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 23680
    },
    {
      "epoch": 9.46104674390731,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 23681
    },
    {
      "epoch": 9.461446264482621,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 23682
    },
    {
      "epoch": 9.46184578505793,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 23683
    },
    {
      "epoch": 9.46224530563324,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 23684
    },
    {
      "epoch": 9.46264482620855,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 23685
    },
    {
      "epoch": 9.463044346783859,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 23686
    },
    {
      "epoch": 9.46344386735917,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 23687
    },
    {
      "epoch": 9.463843387934478,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 23688
    },
    {
      "epoch": 9.464242908509789,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 23689
    },
    {
      "epoch": 9.464642429085098,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 23690
    },
    {
      "epoch": 9.465041949660408,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 23691
    },
    {
      "epoch": 9.465441470235717,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 23692
    },
    {
      "epoch": 9.465840990811026,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 23693
    },
    {
      "epoch": 9.466240511386337,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 23694
    },
    {
      "epoch": 9.466640031961646,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 23695
    },
    {
      "epoch": 9.467039552536956,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 23696
    },
    {
      "epoch": 9.467439073112265,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 23697
    },
    {
      "epoch": 9.467838593687574,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 23698
    },
    {
      "epoch": 9.468238114262885,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 23699
    },
    {
      "epoch": 9.468637634838194,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 23700
    },
    {
      "epoch": 9.469037155413504,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 23701
    },
    {
      "epoch": 9.469436675988813,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 23702
    },
    {
      "epoch": 9.469836196564122,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 23703
    },
    {
      "epoch": 9.470235717139433,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 23704
    },
    {
      "epoch": 9.470635237714742,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 23705
    },
    {
      "epoch": 9.471034758290052,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 23706
    },
    {
      "epoch": 9.471434278865361,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 23707
    },
    {
      "epoch": 9.471833799440672,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 23708
    },
    {
      "epoch": 9.472233320015981,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 23709
    },
    {
      "epoch": 9.47263284059129,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 23710
    },
    {
      "epoch": 9.4730323611666,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 23711
    },
    {
      "epoch": 9.47343188174191,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 23712
    },
    {
      "epoch": 9.47383140231722,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 23713
    },
    {
      "epoch": 9.474230922892529,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 23714
    },
    {
      "epoch": 9.474630443467838,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 23715
    },
    {
      "epoch": 9.475029964043149,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 23716
    },
    {
      "epoch": 9.475429484618457,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 23717
    },
    {
      "epoch": 9.475829005193768,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 23718
    },
    {
      "epoch": 9.476228525769077,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 23719
    },
    {
      "epoch": 9.476628046344386,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 23720
    },
    {
      "epoch": 9.477027566919697,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 23721
    },
    {
      "epoch": 9.477427087495006,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 23722
    },
    {
      "epoch": 9.477826608070316,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 23723
    },
    {
      "epoch": 9.478226128645625,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 23724
    },
    {
      "epoch": 9.478625649220934,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 23725
    },
    {
      "epoch": 9.479025169796245,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 23726
    },
    {
      "epoch": 9.479424690371554,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 23727
    },
    {
      "epoch": 9.479824210946864,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 23728
    },
    {
      "epoch": 9.480223731522173,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 23729
    },
    {
      "epoch": 9.480623252097484,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 23730
    },
    {
      "epoch": 9.481022772672793,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 23731
    },
    {
      "epoch": 9.481422293248102,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 23732
    },
    {
      "epoch": 9.481821813823412,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 23733
    },
    {
      "epoch": 9.482221334398721,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 23734
    },
    {
      "epoch": 9.482620854974032,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0496,
      "step": 23735
    },
    {
      "epoch": 9.48302037554934,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 23736
    },
    {
      "epoch": 9.48341989612465,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 23737
    },
    {
      "epoch": 9.48381941669996,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 23738
    },
    {
      "epoch": 9.48421893727527,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 23739
    },
    {
      "epoch": 9.48461845785058,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 23740
    },
    {
      "epoch": 9.485017978425889,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 23741
    },
    {
      "epoch": 9.485417499001198,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 23742
    },
    {
      "epoch": 9.485817019576508,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 23743
    },
    {
      "epoch": 9.486216540151817,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 23744
    },
    {
      "epoch": 9.486616060727128,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 23745
    },
    {
      "epoch": 9.487015581302437,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 23746
    },
    {
      "epoch": 9.487415101877748,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 23747
    },
    {
      "epoch": 9.487814622453056,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 23748
    },
    {
      "epoch": 9.488214143028365,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 23749
    },
    {
      "epoch": 9.488613663603676,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 23750
    },
    {
      "epoch": 9.489013184178985,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 23751
    },
    {
      "epoch": 9.489412704754296,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 23752
    },
    {
      "epoch": 9.489812225329604,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 23753
    },
    {
      "epoch": 9.490211745904913,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 23754
    },
    {
      "epoch": 9.490611266480224,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 23755
    },
    {
      "epoch": 9.491010787055533,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 23756
    },
    {
      "epoch": 9.491410307630844,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 23757
    },
    {
      "epoch": 9.491809828206152,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 23758
    },
    {
      "epoch": 9.492209348781461,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 23759
    },
    {
      "epoch": 9.492608869356772,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 23760
    },
    {
      "epoch": 9.493008389932081,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 23761
    },
    {
      "epoch": 9.493407910507392,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 23762
    },
    {
      "epoch": 9.4938074310827,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 23763
    },
    {
      "epoch": 9.494206951658011,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 23764
    },
    {
      "epoch": 9.49460647223332,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 23765
    },
    {
      "epoch": 9.495005992808629,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 23766
    },
    {
      "epoch": 9.49540551338394,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 23767
    },
    {
      "epoch": 9.495805033959249,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 23768
    },
    {
      "epoch": 9.49620455453456,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 23769
    },
    {
      "epoch": 9.496604075109868,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 23770
    },
    {
      "epoch": 9.497003595685177,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 23771
    },
    {
      "epoch": 9.497403116260488,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 23772
    },
    {
      "epoch": 9.497802636835797,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 23773
    },
    {
      "epoch": 9.498202157411107,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 23774
    },
    {
      "epoch": 9.498601677986416,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 23775
    },
    {
      "epoch": 9.499001198561725,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 23776
    },
    {
      "epoch": 9.499400719137036,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 23777
    },
    {
      "epoch": 9.499800239712345,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 23778
    },
    {
      "epoch": 9.500199760287655,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 23779
    },
    {
      "epoch": 9.500599280862964,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 23780
    },
    {
      "epoch": 9.500998801438275,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 23781
    },
    {
      "epoch": 9.501398322013584,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 23782
    },
    {
      "epoch": 9.501797842588893,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 23783
    },
    {
      "epoch": 9.502197363164203,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 23784
    },
    {
      "epoch": 9.502596883739512,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 23785
    },
    {
      "epoch": 9.502996404314823,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 23786
    },
    {
      "epoch": 9.503395924890132,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 23787
    },
    {
      "epoch": 9.50379544546544,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 23788
    },
    {
      "epoch": 9.504194966040751,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 23789
    },
    {
      "epoch": 9.50459448661606,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 23790
    },
    {
      "epoch": 9.504994007191371,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 23791
    },
    {
      "epoch": 9.50539352776668,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 23792
    },
    {
      "epoch": 9.505793048341989,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 23793
    },
    {
      "epoch": 9.5061925689173,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 23794
    },
    {
      "epoch": 9.506592089492608,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 23795
    },
    {
      "epoch": 9.506991610067919,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 23796
    },
    {
      "epoch": 9.507391130643228,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 23797
    },
    {
      "epoch": 9.507790651218539,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 23798
    },
    {
      "epoch": 9.508190171793848,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 23799
    },
    {
      "epoch": 9.508589692369156,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 23800
    },
    {
      "epoch": 9.508989212944467,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 23801
    },
    {
      "epoch": 9.509388733519776,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 23802
    },
    {
      "epoch": 9.509788254095087,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 23803
    },
    {
      "epoch": 9.510187774670396,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 23804
    },
    {
      "epoch": 9.510587295245704,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 23805
    },
    {
      "epoch": 9.510986815821015,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 23806
    },
    {
      "epoch": 9.511386336396324,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 23807
    },
    {
      "epoch": 9.511785856971635,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 23808
    },
    {
      "epoch": 9.512185377546944,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 23809
    },
    {
      "epoch": 9.512584898122252,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 23810
    },
    {
      "epoch": 9.512984418697563,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 23811
    },
    {
      "epoch": 9.513383939272872,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 23812
    },
    {
      "epoch": 9.513783459848183,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 23813
    },
    {
      "epoch": 9.514182980423492,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 23814
    },
    {
      "epoch": 9.514582500998802,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 23815
    },
    {
      "epoch": 9.514982021574111,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 23816
    },
    {
      "epoch": 9.51538154214942,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 23817
    },
    {
      "epoch": 9.51578106272473,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 23818
    },
    {
      "epoch": 9.51618058330004,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 23819
    },
    {
      "epoch": 9.51658010387535,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 23820
    },
    {
      "epoch": 9.51697962445066,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 23821
    },
    {
      "epoch": 9.517379145025968,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 23822
    },
    {
      "epoch": 9.517778665601279,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 23823
    },
    {
      "epoch": 9.518178186176588,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 23824
    },
    {
      "epoch": 9.518577706751898,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 23825
    },
    {
      "epoch": 9.518977227327207,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 23826
    },
    {
      "epoch": 9.519376747902516,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 23827
    },
    {
      "epoch": 9.519776268477827,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 23828
    },
    {
      "epoch": 9.520175789053136,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 23829
    },
    {
      "epoch": 9.520575309628446,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 23830
    },
    {
      "epoch": 9.520974830203755,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 23831
    },
    {
      "epoch": 9.521374350779066,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 23832
    },
    {
      "epoch": 9.521773871354375,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 23833
    },
    {
      "epoch": 9.522173391929684,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 23834
    },
    {
      "epoch": 9.522572912504994,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 23835
    },
    {
      "epoch": 9.522972433080303,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 23836
    },
    {
      "epoch": 9.523371953655614,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 23837
    },
    {
      "epoch": 9.523771474230923,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 23838
    },
    {
      "epoch": 9.524170994806232,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 23839
    },
    {
      "epoch": 9.524570515381543,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 23840
    },
    {
      "epoch": 9.524970035956851,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 23841
    },
    {
      "epoch": 9.525369556532162,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 23842
    },
    {
      "epoch": 9.525769077107471,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 23843
    },
    {
      "epoch": 9.52616859768278,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 23844
    },
    {
      "epoch": 9.52656811825809,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 23845
    },
    {
      "epoch": 9.5269676388334,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 23846
    },
    {
      "epoch": 9.52736715940871,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 23847
    },
    {
      "epoch": 9.527766679984019,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 23848
    },
    {
      "epoch": 9.52816620055933,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 23849
    },
    {
      "epoch": 9.528565721134639,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 23850
    },
    {
      "epoch": 9.528965241709948,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 23851
    },
    {
      "epoch": 9.529364762285258,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 23852
    },
    {
      "epoch": 9.529764282860567,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 23853
    },
    {
      "epoch": 9.530163803435878,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 23854
    },
    {
      "epoch": 9.530563324011187,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 23855
    },
    {
      "epoch": 9.530962844586496,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 23856
    },
    {
      "epoch": 9.531362365161806,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 23857
    },
    {
      "epoch": 9.531761885737115,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 23858
    },
    {
      "epoch": 9.532161406312426,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 23859
    },
    {
      "epoch": 9.532560926887735,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 23860
    },
    {
      "epoch": 9.532960447463044,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 23861
    },
    {
      "epoch": 9.533359968038354,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0518,
      "step": 23862
    },
    {
      "epoch": 9.533759488613663,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 23863
    },
    {
      "epoch": 9.534159009188974,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 23864
    },
    {
      "epoch": 9.534558529764283,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 23865
    },
    {
      "epoch": 9.534958050339593,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 23866
    },
    {
      "epoch": 9.535357570914902,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 23867
    },
    {
      "epoch": 9.535757091490211,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 23868
    },
    {
      "epoch": 9.536156612065522,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 23869
    },
    {
      "epoch": 9.53655613264083,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 23870
    },
    {
      "epoch": 9.53695565321614,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 23871
    },
    {
      "epoch": 9.53735517379145,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 23872
    },
    {
      "epoch": 9.53775469436676,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 23873
    },
    {
      "epoch": 9.53815421494207,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 23874
    },
    {
      "epoch": 9.538553735517379,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 23875
    },
    {
      "epoch": 9.53895325609269,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 23876
    },
    {
      "epoch": 9.539352776667998,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 23877
    },
    {
      "epoch": 9.539752297243307,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 23878
    },
    {
      "epoch": 9.540151817818618,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 23879
    },
    {
      "epoch": 9.540551338393927,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 23880
    },
    {
      "epoch": 9.540950858969238,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 23881
    },
    {
      "epoch": 9.541350379544546,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 23882
    },
    {
      "epoch": 9.541749900119855,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 23883
    },
    {
      "epoch": 9.542149420695166,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 23884
    },
    {
      "epoch": 9.542548941270475,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 23885
    },
    {
      "epoch": 9.542948461845786,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 23886
    },
    {
      "epoch": 9.543347982421094,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 23887
    },
    {
      "epoch": 9.543747502996403,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 23888
    },
    {
      "epoch": 9.544147023571714,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 23889
    },
    {
      "epoch": 9.544546544147023,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 23890
    },
    {
      "epoch": 9.544946064722334,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 23891
    },
    {
      "epoch": 9.545345585297643,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 23892
    },
    {
      "epoch": 9.545745105872953,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 23893
    },
    {
      "epoch": 9.546144626448262,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 23894
    },
    {
      "epoch": 9.546544147023571,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 23895
    },
    {
      "epoch": 9.546943667598882,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 23896
    },
    {
      "epoch": 9.54734318817419,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 23897
    },
    {
      "epoch": 9.547742708749501,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 23898
    },
    {
      "epoch": 9.54814222932481,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 23899
    },
    {
      "epoch": 9.548541749900119,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 23900
    },
    {
      "epoch": 9.54894127047543,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 23901
    },
    {
      "epoch": 9.549340791050739,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 23902
    },
    {
      "epoch": 9.54974031162605,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 23903
    },
    {
      "epoch": 9.550139832201358,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 23904
    },
    {
      "epoch": 9.550539352776667,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 23905
    },
    {
      "epoch": 9.550938873351978,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 23906
    },
    {
      "epoch": 9.551338393927287,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 23907
    },
    {
      "epoch": 9.551737914502597,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 23908
    },
    {
      "epoch": 9.552137435077906,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 23909
    },
    {
      "epoch": 9.552536955653217,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 23910
    },
    {
      "epoch": 9.552936476228526,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 23911
    },
    {
      "epoch": 9.553335996803835,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 23912
    },
    {
      "epoch": 9.553735517379145,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 23913
    },
    {
      "epoch": 9.554135037954454,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 23914
    },
    {
      "epoch": 9.554534558529765,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 23915
    },
    {
      "epoch": 9.554934079105074,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 23916
    },
    {
      "epoch": 9.555333599680383,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 23917
    },
    {
      "epoch": 9.555733120255693,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 23918
    },
    {
      "epoch": 9.556132640831002,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 23919
    },
    {
      "epoch": 9.556532161406313,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 23920
    },
    {
      "epoch": 9.556931681981622,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 23921
    },
    {
      "epoch": 9.55733120255693,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 23922
    },
    {
      "epoch": 9.557730723132241,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 23923
    },
    {
      "epoch": 9.55813024370755,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 23924
    },
    {
      "epoch": 9.558529764282861,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 23925
    },
    {
      "epoch": 9.55892928485817,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 23926
    },
    {
      "epoch": 9.55932880543348,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 23927
    },
    {
      "epoch": 9.55972832600879,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 23928
    },
    {
      "epoch": 9.560127846584098,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 23929
    },
    {
      "epoch": 9.560527367159409,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 23930
    },
    {
      "epoch": 9.560926887734718,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 23931
    },
    {
      "epoch": 9.561326408310029,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 23932
    },
    {
      "epoch": 9.561725928885338,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 23933
    },
    {
      "epoch": 9.562125449460646,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 23934
    },
    {
      "epoch": 9.562524970035957,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0508,
      "step": 23935
    },
    {
      "epoch": 9.562924490611266,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 23936
    },
    {
      "epoch": 9.563324011186577,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 23937
    },
    {
      "epoch": 9.563723531761886,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 23938
    },
    {
      "epoch": 9.564123052337195,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 23939
    },
    {
      "epoch": 9.564522572912505,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 23940
    },
    {
      "epoch": 9.564922093487814,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 23941
    },
    {
      "epoch": 9.565321614063125,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 23942
    },
    {
      "epoch": 9.565721134638434,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 23943
    },
    {
      "epoch": 9.566120655213744,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 23944
    },
    {
      "epoch": 9.566520175789053,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 23945
    },
    {
      "epoch": 9.566919696364362,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 23946
    },
    {
      "epoch": 9.567319216939673,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 23947
    },
    {
      "epoch": 9.567718737514982,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 23948
    },
    {
      "epoch": 9.568118258090292,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 23949
    },
    {
      "epoch": 9.568517778665601,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 23950
    },
    {
      "epoch": 9.56891729924091,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 23951
    },
    {
      "epoch": 9.56931681981622,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 23952
    },
    {
      "epoch": 9.56971634039153,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 23953
    },
    {
      "epoch": 9.57011586096684,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 23954
    },
    {
      "epoch": 9.57051538154215,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 23955
    },
    {
      "epoch": 9.570914902117458,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 23956
    },
    {
      "epoch": 9.571314422692769,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 23957
    },
    {
      "epoch": 9.571713943268078,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 23958
    },
    {
      "epoch": 9.572113463843388,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 23959
    },
    {
      "epoch": 9.572512984418697,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 23960
    },
    {
      "epoch": 9.572912504994008,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 23961
    },
    {
      "epoch": 9.573312025569317,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 23962
    },
    {
      "epoch": 9.573711546144626,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 23963
    },
    {
      "epoch": 9.574111066719937,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 23964
    },
    {
      "epoch": 9.574510587295245,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 23965
    },
    {
      "epoch": 9.574910107870556,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 23966
    },
    {
      "epoch": 9.575309628445865,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 23967
    },
    {
      "epoch": 9.575709149021174,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 23968
    },
    {
      "epoch": 9.576108669596485,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 23969
    },
    {
      "epoch": 9.576508190171793,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 23970
    },
    {
      "epoch": 9.576907710747104,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 23971
    },
    {
      "epoch": 9.577307231322413,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 23972
    },
    {
      "epoch": 9.577706751897722,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 23973
    },
    {
      "epoch": 9.578106272473033,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 23974
    },
    {
      "epoch": 9.578505793048341,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 23975
    },
    {
      "epoch": 9.578905313623652,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 23976
    },
    {
      "epoch": 9.579304834198961,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 23977
    },
    {
      "epoch": 9.579704354774272,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 23978
    },
    {
      "epoch": 9.58010387534958,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 23979
    },
    {
      "epoch": 9.58050339592489,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 23980
    },
    {
      "epoch": 9.5809029165002,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 23981
    },
    {
      "epoch": 9.581302437075509,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 23982
    },
    {
      "epoch": 9.58170195765082,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 23983
    },
    {
      "epoch": 9.582101478226129,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 23984
    },
    {
      "epoch": 9.582500998801438,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 23985
    },
    {
      "epoch": 9.582900519376748,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 23986
    },
    {
      "epoch": 9.583300039952057,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 23987
    },
    {
      "epoch": 9.583699560527368,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 23988
    },
    {
      "epoch": 9.584099081102677,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 23989
    },
    {
      "epoch": 9.584498601677986,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 23990
    },
    {
      "epoch": 9.584898122253296,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 23991
    },
    {
      "epoch": 9.585297642828605,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 23992
    },
    {
      "epoch": 9.585697163403916,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 23993
    },
    {
      "epoch": 9.586096683979225,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 23994
    },
    {
      "epoch": 9.586496204554535,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 23995
    },
    {
      "epoch": 9.586895725129844,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 23996
    },
    {
      "epoch": 9.587295245705153,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 23997
    },
    {
      "epoch": 9.587694766280464,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 23998
    },
    {
      "epoch": 9.588094286855773,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 23999
    },
    {
      "epoch": 9.588493807431083,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 24000
    },
    {
      "epoch": 9.588893328006392,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 24001
    },
    {
      "epoch": 9.589292848581701,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 24002
    },
    {
      "epoch": 9.589692369157012,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 24003
    },
    {
      "epoch": 9.59009188973232,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 24004
    },
    {
      "epoch": 9.590491410307632,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 24005
    },
    {
      "epoch": 9.59089093088294,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 24006
    },
    {
      "epoch": 9.59129045145825,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 24007
    },
    {
      "epoch": 9.59168997203356,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 24008
    },
    {
      "epoch": 9.592089492608869,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 24009
    },
    {
      "epoch": 9.59248901318418,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 24010
    },
    {
      "epoch": 9.592888533759488,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 24011
    },
    {
      "epoch": 9.5932880543348,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 24012
    },
    {
      "epoch": 9.593687574910108,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 24013
    },
    {
      "epoch": 9.594087095485417,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 24014
    },
    {
      "epoch": 9.594486616060728,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 24015
    },
    {
      "epoch": 9.594886136636037,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 24016
    },
    {
      "epoch": 9.595285657211347,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 24017
    },
    {
      "epoch": 9.595685177786656,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 24018
    },
    {
      "epoch": 9.596084698361965,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 24019
    },
    {
      "epoch": 9.596484218937276,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 24020
    },
    {
      "epoch": 9.596883739512585,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 24021
    },
    {
      "epoch": 9.597283260087895,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 24022
    },
    {
      "epoch": 9.597682780663204,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 24023
    },
    {
      "epoch": 9.598082301238513,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 24024
    },
    {
      "epoch": 9.598481821813824,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 24025
    },
    {
      "epoch": 9.598881342389133,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 24026
    },
    {
      "epoch": 9.599280862964443,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 24027
    },
    {
      "epoch": 9.599680383539752,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 24028
    },
    {
      "epoch": 9.600079904115063,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 24029
    },
    {
      "epoch": 9.600479424690372,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 24030
    },
    {
      "epoch": 9.60087894526568,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 24031
    },
    {
      "epoch": 9.601278465840991,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 24032
    },
    {
      "epoch": 9.6016779864163,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 24033
    },
    {
      "epoch": 9.602077506991611,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 24034
    },
    {
      "epoch": 9.60247702756692,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 24035
    },
    {
      "epoch": 9.602876548142229,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 24036
    },
    {
      "epoch": 9.60327606871754,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 24037
    },
    {
      "epoch": 9.603675589292848,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 24038
    },
    {
      "epoch": 9.604075109868159,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 24039
    },
    {
      "epoch": 9.604474630443468,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 24040
    },
    {
      "epoch": 9.604874151018777,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 24041
    },
    {
      "epoch": 9.605273671594087,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 24042
    },
    {
      "epoch": 9.605673192169396,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 24043
    },
    {
      "epoch": 9.606072712744707,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 24044
    },
    {
      "epoch": 9.606472233320016,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 24045
    },
    {
      "epoch": 9.606871753895327,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 24046
    },
    {
      "epoch": 9.607271274470635,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 24047
    },
    {
      "epoch": 9.607670795045944,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 24048
    },
    {
      "epoch": 9.608070315621255,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 24049
    },
    {
      "epoch": 9.608469836196564,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 24050
    },
    {
      "epoch": 9.608869356771873,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 24051
    },
    {
      "epoch": 9.609268877347183,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 24052
    },
    {
      "epoch": 9.609668397922492,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 24053
    },
    {
      "epoch": 9.610067918497803,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 24054
    },
    {
      "epoch": 9.610467439073112,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 24055
    },
    {
      "epoch": 9.610866959648423,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 24056
    },
    {
      "epoch": 9.611266480223732,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 24057
    },
    {
      "epoch": 9.61166600079904,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 24058
    },
    {
      "epoch": 9.612065521374351,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 24059
    },
    {
      "epoch": 9.61246504194966,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 24060
    },
    {
      "epoch": 9.61286456252497,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 24061
    },
    {
      "epoch": 9.61326408310028,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 24062
    },
    {
      "epoch": 9.613663603675588,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 24063
    },
    {
      "epoch": 9.6140631242509,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 24064
    },
    {
      "epoch": 9.614462644826208,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 24065
    },
    {
      "epoch": 9.614862165401519,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 24066
    },
    {
      "epoch": 9.615261685976828,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 24067
    },
    {
      "epoch": 9.615661206552137,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 24068
    },
    {
      "epoch": 9.616060727127447,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 24069
    },
    {
      "epoch": 9.616460247702756,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 24070
    },
    {
      "epoch": 9.616859768278067,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 24071
    },
    {
      "epoch": 9.617259288853376,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 24072
    },
    {
      "epoch": 9.617658809428686,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 24073
    },
    {
      "epoch": 9.618058330003995,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 24074
    },
    {
      "epoch": 9.618457850579304,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 24075
    },
    {
      "epoch": 9.618857371154615,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 24076
    },
    {
      "epoch": 9.619256891729924,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 24077
    },
    {
      "epoch": 9.619656412305234,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 24078
    },
    {
      "epoch": 9.620055932880543,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 24079
    },
    {
      "epoch": 9.620455453455852,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 24080
    },
    {
      "epoch": 9.620854974031163,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 24081
    },
    {
      "epoch": 9.621254494606472,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 24082
    },
    {
      "epoch": 9.621654015181782,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 24083
    },
    {
      "epoch": 9.622053535757091,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 24084
    },
    {
      "epoch": 9.6224530563324,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 24085
    },
    {
      "epoch": 9.622852576907711,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 24086
    },
    {
      "epoch": 9.62325209748302,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 24087
    },
    {
      "epoch": 9.62365161805833,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 24088
    },
    {
      "epoch": 9.62405113863364,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 24089
    },
    {
      "epoch": 9.62445065920895,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 24090
    },
    {
      "epoch": 9.624850179784259,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 24091
    },
    {
      "epoch": 9.625249700359568,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 24092
    },
    {
      "epoch": 9.625649220934879,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 24093
    },
    {
      "epoch": 9.626048741510187,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 24094
    },
    {
      "epoch": 9.626448262085498,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 24095
    },
    {
      "epoch": 9.626847782660807,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 24096
    },
    {
      "epoch": 9.627247303236116,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 24097
    },
    {
      "epoch": 9.627646823811427,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 24098
    },
    {
      "epoch": 9.628046344386735,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 24099
    },
    {
      "epoch": 9.628445864962046,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 24100
    },
    {
      "epoch": 9.628845385537355,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 24101
    },
    {
      "epoch": 9.629244906112664,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 24102
    },
    {
      "epoch": 9.629644426687975,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 24103
    },
    {
      "epoch": 9.630043947263283,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 24104
    },
    {
      "epoch": 9.630443467838594,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 24105
    },
    {
      "epoch": 9.630842988413903,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 24106
    },
    {
      "epoch": 9.631242508989214,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 24107
    },
    {
      "epoch": 9.631642029564523,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 24108
    },
    {
      "epoch": 9.632041550139832,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 24109
    },
    {
      "epoch": 9.632441070715142,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 24110
    },
    {
      "epoch": 9.632840591290451,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 24111
    },
    {
      "epoch": 9.633240111865762,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 24112
    },
    {
      "epoch": 9.63363963244107,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 24113
    },
    {
      "epoch": 9.63403915301638,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 24114
    },
    {
      "epoch": 9.63443867359169,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 24115
    },
    {
      "epoch": 9.634838194167,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 24116
    },
    {
      "epoch": 9.63523771474231,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 24117
    },
    {
      "epoch": 9.635637235317619,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 24118
    },
    {
      "epoch": 9.636036755892928,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 24119
    },
    {
      "epoch": 9.636436276468238,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 24120
    },
    {
      "epoch": 9.636835797043547,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 24121
    },
    {
      "epoch": 9.637235317618858,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 24122
    },
    {
      "epoch": 9.637634838194167,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 24123
    },
    {
      "epoch": 9.638034358769477,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 24124
    },
    {
      "epoch": 9.638433879344786,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.012,
      "step": 24125
    },
    {
      "epoch": 9.638833399920095,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 24126
    },
    {
      "epoch": 9.639232920495406,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 24127
    },
    {
      "epoch": 9.639632441070715,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 24128
    },
    {
      "epoch": 9.640031961646025,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 24129
    },
    {
      "epoch": 9.640431482221334,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 24130
    },
    {
      "epoch": 9.640831002796643,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 24131
    },
    {
      "epoch": 9.641230523371954,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 24132
    },
    {
      "epoch": 9.641630043947263,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 24133
    },
    {
      "epoch": 9.642029564522574,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 24134
    },
    {
      "epoch": 9.642429085097882,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 24135
    },
    {
      "epoch": 9.642828605673191,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 24136
    },
    {
      "epoch": 9.643228126248502,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 24137
    },
    {
      "epoch": 9.643627646823811,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 24138
    },
    {
      "epoch": 9.644027167399122,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 24139
    },
    {
      "epoch": 9.64442668797443,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 24140
    },
    {
      "epoch": 9.644826208549741,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 24141
    },
    {
      "epoch": 9.64522572912505,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 24142
    },
    {
      "epoch": 9.645625249700359,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 24143
    },
    {
      "epoch": 9.64602477027567,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 24144
    },
    {
      "epoch": 9.646424290850979,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 24145
    },
    {
      "epoch": 9.64682381142629,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 24146
    },
    {
      "epoch": 9.647223332001598,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 24147
    },
    {
      "epoch": 9.647622852576907,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 24148
    },
    {
      "epoch": 9.648022373152218,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 24149
    },
    {
      "epoch": 9.648421893727527,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 24150
    },
    {
      "epoch": 9.648821414302837,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 24151
    },
    {
      "epoch": 9.649220934878146,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 24152
    },
    {
      "epoch": 9.649620455453455,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 24153
    },
    {
      "epoch": 9.650019976028766,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 24154
    },
    {
      "epoch": 9.650419496604075,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 24155
    },
    {
      "epoch": 9.650819017179385,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 24156
    },
    {
      "epoch": 9.651218537754694,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 24157
    },
    {
      "epoch": 9.651618058330005,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 24158
    },
    {
      "epoch": 9.652017578905314,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 24159
    },
    {
      "epoch": 9.652417099480623,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 24160
    },
    {
      "epoch": 9.652816620055933,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 24161
    },
    {
      "epoch": 9.653216140631242,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 24162
    },
    {
      "epoch": 9.653615661206553,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 24163
    },
    {
      "epoch": 9.654015181781862,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 24164
    },
    {
      "epoch": 9.65441470235717,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 24165
    },
    {
      "epoch": 9.654814222932481,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 24166
    },
    {
      "epoch": 9.65521374350779,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 24167
    },
    {
      "epoch": 9.655613264083101,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 24168
    },
    {
      "epoch": 9.65601278465841,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 24169
    },
    {
      "epoch": 9.656412305233719,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 24170
    },
    {
      "epoch": 9.65681182580903,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 24171
    },
    {
      "epoch": 9.657211346384338,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 24172
    },
    {
      "epoch": 9.657610866959649,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 24173
    },
    {
      "epoch": 9.658010387534958,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 24174
    },
    {
      "epoch": 9.658409908110269,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 24175
    },
    {
      "epoch": 9.658809428685577,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 24176
    },
    {
      "epoch": 9.659208949260886,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 24177
    },
    {
      "epoch": 9.659608469836197,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 24178
    },
    {
      "epoch": 9.660007990411506,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 24179
    },
    {
      "epoch": 9.660407510986817,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 24180
    },
    {
      "epoch": 9.660807031562126,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 24181
    },
    {
      "epoch": 9.661206552137434,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 24182
    },
    {
      "epoch": 9.661606072712745,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 24183
    },
    {
      "epoch": 9.662005593288054,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 24184
    },
    {
      "epoch": 9.662405113863365,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 24185
    },
    {
      "epoch": 9.662804634438674,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 24186
    },
    {
      "epoch": 9.663204155013982,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 24187
    },
    {
      "epoch": 9.663603675589293,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 24188
    },
    {
      "epoch": 9.664003196164602,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 24189
    },
    {
      "epoch": 9.664402716739913,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 24190
    },
    {
      "epoch": 9.664802237315222,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 24191
    },
    {
      "epoch": 9.665201757890532,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 24192
    },
    {
      "epoch": 9.665601278465841,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 24193
    },
    {
      "epoch": 9.66600079904115,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 24194
    },
    {
      "epoch": 9.66640031961646,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 24195
    },
    {
      "epoch": 9.66679984019177,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 24196
    },
    {
      "epoch": 9.66719936076708,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 24197
    },
    {
      "epoch": 9.66759888134239,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 24198
    },
    {
      "epoch": 9.667998401917698,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 24199
    },
    {
      "epoch": 9.668397922493009,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 24200
    },
    {
      "epoch": 9.668797443068318,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 24201
    },
    {
      "epoch": 9.669196963643628,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 24202
    },
    {
      "epoch": 9.669596484218937,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 24203
    },
    {
      "epoch": 9.669996004794246,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 24204
    },
    {
      "epoch": 9.670395525369557,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 24205
    },
    {
      "epoch": 9.670795045944866,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 24206
    },
    {
      "epoch": 9.671194566520176,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 24207
    },
    {
      "epoch": 9.671594087095485,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 24208
    },
    {
      "epoch": 9.671993607670796,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 24209
    },
    {
      "epoch": 9.672393128246105,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 24210
    },
    {
      "epoch": 9.672792648821414,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 24211
    },
    {
      "epoch": 9.673192169396724,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 24212
    },
    {
      "epoch": 9.673591689972033,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 24213
    },
    {
      "epoch": 9.673991210547344,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 24214
    },
    {
      "epoch": 9.674390731122653,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 24215
    },
    {
      "epoch": 9.674790251697962,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 24216
    },
    {
      "epoch": 9.675189772273272,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 24217
    },
    {
      "epoch": 9.675589292848581,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 24218
    },
    {
      "epoch": 9.675988813423892,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 24219
    },
    {
      "epoch": 9.676388333999201,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 24220
    },
    {
      "epoch": 9.67678785457451,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 24221
    },
    {
      "epoch": 9.67718737514982,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 24222
    },
    {
      "epoch": 9.67758689572513,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 24223
    },
    {
      "epoch": 9.67798641630044,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 24224
    },
    {
      "epoch": 9.678385936875749,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 24225
    },
    {
      "epoch": 9.67878545745106,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 24226
    },
    {
      "epoch": 9.679184978026369,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 24227
    },
    {
      "epoch": 9.679584498601677,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 24228
    },
    {
      "epoch": 9.679984019176988,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 24229
    },
    {
      "epoch": 9.680383539752297,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 24230
    },
    {
      "epoch": 9.680783060327606,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 24231
    },
    {
      "epoch": 9.681182580902917,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 24232
    },
    {
      "epoch": 9.681582101478226,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 24233
    },
    {
      "epoch": 9.681981622053536,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 24234
    },
    {
      "epoch": 9.682381142628845,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 24235
    },
    {
      "epoch": 9.682780663204156,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 24236
    },
    {
      "epoch": 9.683180183779465,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 24237
    },
    {
      "epoch": 9.683579704354774,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 24238
    },
    {
      "epoch": 9.683979224930084,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 24239
    },
    {
      "epoch": 9.684378745505393,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 24240
    },
    {
      "epoch": 9.684778266080704,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 24241
    },
    {
      "epoch": 9.685177786656013,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 24242
    },
    {
      "epoch": 9.685577307231322,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 24243
    },
    {
      "epoch": 9.685976827806632,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 24244
    },
    {
      "epoch": 9.686376348381941,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 24245
    },
    {
      "epoch": 9.686775868957252,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 24246
    },
    {
      "epoch": 9.68717538953256,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 24247
    },
    {
      "epoch": 9.68757491010787,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 24248
    },
    {
      "epoch": 9.68797443068318,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 24249
    },
    {
      "epoch": 9.68837395125849,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 24250
    },
    {
      "epoch": 9.6887734718338,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 24251
    },
    {
      "epoch": 9.689172992409109,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 24252
    },
    {
      "epoch": 9.68957251298442,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 24253
    },
    {
      "epoch": 9.689972033559728,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 24254
    },
    {
      "epoch": 9.690371554135037,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 24255
    },
    {
      "epoch": 9.690771074710348,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 24256
    },
    {
      "epoch": 9.691170595285657,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 24257
    },
    {
      "epoch": 9.691570115860968,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 24258
    },
    {
      "epoch": 9.691969636436276,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 24259
    },
    {
      "epoch": 9.692369157011585,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 24260
    },
    {
      "epoch": 9.692768677586896,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 24261
    },
    {
      "epoch": 9.693168198162205,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 24262
    },
    {
      "epoch": 9.693567718737516,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 24263
    },
    {
      "epoch": 9.693967239312824,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 24264
    },
    {
      "epoch": 9.694366759888133,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 24265
    },
    {
      "epoch": 9.694766280463444,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 24266
    },
    {
      "epoch": 9.695165801038753,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 24267
    },
    {
      "epoch": 9.695565321614064,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 24268
    },
    {
      "epoch": 9.695964842189372,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 24269
    },
    {
      "epoch": 9.696364362764683,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 24270
    },
    {
      "epoch": 9.696763883339992,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 24271
    },
    {
      "epoch": 9.697163403915301,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 24272
    },
    {
      "epoch": 9.697562924490612,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 24273
    },
    {
      "epoch": 9.69796244506592,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 24274
    },
    {
      "epoch": 9.698361965641231,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 24275
    },
    {
      "epoch": 9.69876148621654,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 24276
    },
    {
      "epoch": 9.699161006791849,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 24277
    },
    {
      "epoch": 9.69956052736716,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 24278
    },
    {
      "epoch": 9.699960047942469,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 24279
    },
    {
      "epoch": 9.70035956851778,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 24280
    },
    {
      "epoch": 9.700759089093088,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 24281
    },
    {
      "epoch": 9.701158609668397,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 24282
    },
    {
      "epoch": 9.701558130243708,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 24283
    },
    {
      "epoch": 9.701957650819017,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 24284
    },
    {
      "epoch": 9.702357171394327,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 24285
    },
    {
      "epoch": 9.702756691969636,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 24286
    },
    {
      "epoch": 9.703156212544947,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 24287
    },
    {
      "epoch": 9.703555733120256,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 24288
    },
    {
      "epoch": 9.703955253695565,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 24289
    },
    {
      "epoch": 9.704354774270875,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 24290
    },
    {
      "epoch": 9.704754294846184,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 24291
    },
    {
      "epoch": 9.705153815421495,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 24292
    },
    {
      "epoch": 9.705553335996804,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 24293
    },
    {
      "epoch": 9.705952856572113,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 24294
    },
    {
      "epoch": 9.706352377147423,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 24295
    },
    {
      "epoch": 9.706751897722732,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 24296
    },
    {
      "epoch": 9.707151418298043,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 24297
    },
    {
      "epoch": 9.707550938873352,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 24298
    },
    {
      "epoch": 9.70795045944866,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 24299
    },
    {
      "epoch": 9.708349980023971,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 24300
    },
    {
      "epoch": 9.70874950059928,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 24301
    },
    {
      "epoch": 9.709149021174591,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 24302
    },
    {
      "epoch": 9.7095485417499,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 24303
    },
    {
      "epoch": 9.70994806232521,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 24304
    },
    {
      "epoch": 9.71034758290052,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 24305
    },
    {
      "epoch": 9.710747103475828,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 24306
    },
    {
      "epoch": 9.711146624051139,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 24307
    },
    {
      "epoch": 9.711546144626448,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 24308
    },
    {
      "epoch": 9.711945665201759,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 24309
    },
    {
      "epoch": 9.712345185777068,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 24310
    },
    {
      "epoch": 9.712744706352376,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 24311
    },
    {
      "epoch": 9.713144226927687,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 24312
    },
    {
      "epoch": 9.713543747502996,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 24313
    },
    {
      "epoch": 9.713943268078307,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 24314
    },
    {
      "epoch": 9.714342788653616,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 24315
    },
    {
      "epoch": 9.714742309228924,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 24316
    },
    {
      "epoch": 9.715141829804235,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 24317
    },
    {
      "epoch": 9.715541350379544,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 24318
    },
    {
      "epoch": 9.715940870954855,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 24319
    },
    {
      "epoch": 9.716340391530164,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 24320
    },
    {
      "epoch": 9.716739912105474,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 24321
    },
    {
      "epoch": 9.717139432680783,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 24322
    },
    {
      "epoch": 9.717538953256092,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 24323
    },
    {
      "epoch": 9.717938473831403,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 24324
    },
    {
      "epoch": 9.718337994406712,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 24325
    },
    {
      "epoch": 9.718737514982022,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 24326
    },
    {
      "epoch": 9.719137035557331,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 24327
    },
    {
      "epoch": 9.71953655613264,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 24328
    },
    {
      "epoch": 9.71993607670795,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 24329
    },
    {
      "epoch": 9.72033559728326,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 24330
    },
    {
      "epoch": 9.72073511785857,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 24331
    },
    {
      "epoch": 9.72113463843388,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 24332
    },
    {
      "epoch": 9.721534159009188,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 24333
    },
    {
      "epoch": 9.721933679584499,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 24334
    },
    {
      "epoch": 9.722333200159808,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 24335
    },
    {
      "epoch": 9.722732720735118,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 24336
    },
    {
      "epoch": 9.723132241310427,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 24337
    },
    {
      "epoch": 9.723531761885738,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 24338
    },
    {
      "epoch": 9.723931282461047,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 24339
    },
    {
      "epoch": 9.724330803036356,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 24340
    },
    {
      "epoch": 9.724730323611666,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 24341
    },
    {
      "epoch": 9.725129844186975,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 24342
    },
    {
      "epoch": 9.725529364762286,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 24343
    },
    {
      "epoch": 9.725928885337595,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 24344
    },
    {
      "epoch": 9.726328405912904,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 24345
    },
    {
      "epoch": 9.726727926488214,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 24346
    },
    {
      "epoch": 9.727127447063523,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 24347
    },
    {
      "epoch": 9.727526967638834,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 24348
    },
    {
      "epoch": 9.727926488214143,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 24349
    },
    {
      "epoch": 9.728326008789452,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 24350
    },
    {
      "epoch": 9.728725529364763,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 24351
    },
    {
      "epoch": 9.729125049940071,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 24352
    },
    {
      "epoch": 9.729524570515382,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 24353
    },
    {
      "epoch": 9.729924091090691,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 24354
    },
    {
      "epoch": 9.730323611666002,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 24355
    },
    {
      "epoch": 9.73072313224131,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 24356
    },
    {
      "epoch": 9.73112265281662,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 24357
    },
    {
      "epoch": 9.73152217339193,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 24358
    },
    {
      "epoch": 9.731921693967239,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 24359
    },
    {
      "epoch": 9.73232121454255,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 24360
    },
    {
      "epoch": 9.732720735117859,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 24361
    },
    {
      "epoch": 9.733120255693168,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 24362
    },
    {
      "epoch": 9.733519776268478,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 24363
    },
    {
      "epoch": 9.733919296843787,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 24364
    },
    {
      "epoch": 9.734318817419098,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 24365
    },
    {
      "epoch": 9.734718337994407,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 24366
    },
    {
      "epoch": 9.735117858569716,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 24367
    },
    {
      "epoch": 9.735517379145026,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 24368
    },
    {
      "epoch": 9.735916899720335,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 24369
    },
    {
      "epoch": 9.736316420295646,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 24370
    },
    {
      "epoch": 9.736715940870955,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 24371
    },
    {
      "epoch": 9.737115461446265,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 24372
    },
    {
      "epoch": 9.737514982021574,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 24373
    },
    {
      "epoch": 9.737914502596883,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 24374
    },
    {
      "epoch": 9.738314023172194,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 24375
    },
    {
      "epoch": 9.738713543747503,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 24376
    },
    {
      "epoch": 9.739113064322813,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 24377
    },
    {
      "epoch": 9.739512584898122,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 24378
    },
    {
      "epoch": 9.739912105473431,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 24379
    },
    {
      "epoch": 9.740311626048742,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 24380
    },
    {
      "epoch": 9.74071114662405,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 24381
    },
    {
      "epoch": 9.741110667199361,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 24382
    },
    {
      "epoch": 9.74151018777467,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 24383
    },
    {
      "epoch": 9.74190970834998,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 24384
    },
    {
      "epoch": 9.74230922892529,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 24385
    },
    {
      "epoch": 9.742708749500599,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 24386
    },
    {
      "epoch": 9.74310827007591,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 24387
    },
    {
      "epoch": 9.743507790651218,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 24388
    },
    {
      "epoch": 9.743907311226529,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 24389
    },
    {
      "epoch": 9.744306831801838,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 24390
    },
    {
      "epoch": 9.744706352377147,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 24391
    },
    {
      "epoch": 9.745105872952458,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 24392
    },
    {
      "epoch": 9.745505393527766,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 24393
    },
    {
      "epoch": 9.745904914103077,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 24394
    },
    {
      "epoch": 9.746304434678386,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 24395
    },
    {
      "epoch": 9.746703955253695,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 24396
    },
    {
      "epoch": 9.747103475829006,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 24397
    },
    {
      "epoch": 9.747502996404315,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 24398
    },
    {
      "epoch": 9.747902516979625,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 24399
    },
    {
      "epoch": 9.748302037554934,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 24400
    },
    {
      "epoch": 9.748701558130243,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 24401
    },
    {
      "epoch": 9.749101078705554,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 24402
    },
    {
      "epoch": 9.749500599280863,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 24403
    },
    {
      "epoch": 9.749900119856173,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 24404
    },
    {
      "epoch": 9.750299640431482,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 24405
    },
    {
      "epoch": 9.750699161006793,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 24406
    },
    {
      "epoch": 9.751098681582102,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 24407
    },
    {
      "epoch": 9.75149820215741,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 24408
    },
    {
      "epoch": 9.751897722732721,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 24409
    },
    {
      "epoch": 9.75229724330803,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 24410
    },
    {
      "epoch": 9.752696763883339,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 24411
    },
    {
      "epoch": 9.75309628445865,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 24412
    },
    {
      "epoch": 9.753495805033959,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 24413
    },
    {
      "epoch": 9.75389532560927,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 24414
    },
    {
      "epoch": 9.754294846184578,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 24415
    },
    {
      "epoch": 9.754694366759889,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 24416
    },
    {
      "epoch": 9.755093887335198,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 24417
    },
    {
      "epoch": 9.755493407910507,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 24418
    },
    {
      "epoch": 9.755892928485817,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 24419
    },
    {
      "epoch": 9.756292449061126,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 24420
    },
    {
      "epoch": 9.756691969636437,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 24421
    },
    {
      "epoch": 9.757091490211746,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 24422
    },
    {
      "epoch": 9.757491010787055,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 24423
    },
    {
      "epoch": 9.757890531362365,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 24424
    },
    {
      "epoch": 9.758290051937674,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 24425
    },
    {
      "epoch": 9.758689572512985,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 24426
    },
    {
      "epoch": 9.759089093088294,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 24427
    },
    {
      "epoch": 9.759488613663603,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 24428
    },
    {
      "epoch": 9.759888134238913,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 24429
    },
    {
      "epoch": 9.760287654814222,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 24430
    },
    {
      "epoch": 9.760687175389533,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 24431
    },
    {
      "epoch": 9.761086695964842,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 24432
    },
    {
      "epoch": 9.761486216540153,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 24433
    },
    {
      "epoch": 9.761885737115461,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 24434
    },
    {
      "epoch": 9.76228525769077,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 24435
    },
    {
      "epoch": 9.762684778266081,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 24436
    },
    {
      "epoch": 9.76308429884139,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 24437
    },
    {
      "epoch": 9.7634838194167,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 24438
    },
    {
      "epoch": 9.76388333999201,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 24439
    },
    {
      "epoch": 9.764282860567318,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 24440
    },
    {
      "epoch": 9.764682381142629,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 24441
    },
    {
      "epoch": 9.765081901717938,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 24442
    },
    {
      "epoch": 9.765481422293249,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 24443
    },
    {
      "epoch": 9.765880942868558,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 24444
    },
    {
      "epoch": 9.766280463443866,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 24445
    },
    {
      "epoch": 9.766679984019177,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 24446
    },
    {
      "epoch": 9.767079504594486,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 24447
    },
    {
      "epoch": 9.767479025169797,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 24448
    },
    {
      "epoch": 9.767878545745106,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 24449
    },
    {
      "epoch": 9.768278066320416,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 24450
    },
    {
      "epoch": 9.768677586895725,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 24451
    },
    {
      "epoch": 9.769077107471034,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 24452
    },
    {
      "epoch": 9.769476628046345,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 24453
    },
    {
      "epoch": 9.769876148621654,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 24454
    },
    {
      "epoch": 9.770275669196964,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 24455
    },
    {
      "epoch": 9.770675189772273,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 24456
    },
    {
      "epoch": 9.771074710347582,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0487,
      "step": 24457
    },
    {
      "epoch": 9.771474230922893,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 24458
    },
    {
      "epoch": 9.771873751498202,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 24459
    },
    {
      "epoch": 9.772273272073512,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 24460
    },
    {
      "epoch": 9.772672792648821,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 24461
    },
    {
      "epoch": 9.77307231322413,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 24462
    },
    {
      "epoch": 9.77347183379944,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 24463
    },
    {
      "epoch": 9.77387135437475,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 24464
    },
    {
      "epoch": 9.77427087495006,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 24465
    },
    {
      "epoch": 9.77467039552537,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 24466
    },
    {
      "epoch": 9.77506991610068,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 24467
    },
    {
      "epoch": 9.775469436675989,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 24468
    },
    {
      "epoch": 9.775868957251298,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 24469
    },
    {
      "epoch": 9.776268477826608,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 24470
    },
    {
      "epoch": 9.776667998401917,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 24471
    },
    {
      "epoch": 9.777067518977228,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 24472
    },
    {
      "epoch": 9.777467039552537,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 24473
    },
    {
      "epoch": 9.777866560127846,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 24474
    },
    {
      "epoch": 9.778266080703157,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 24475
    },
    {
      "epoch": 9.778665601278465,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 24476
    },
    {
      "epoch": 9.779065121853776,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 24477
    },
    {
      "epoch": 9.779464642429085,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 24478
    },
    {
      "epoch": 9.779864163004394,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 24479
    },
    {
      "epoch": 9.780263683579705,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 24480
    },
    {
      "epoch": 9.780663204155013,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 24481
    },
    {
      "epoch": 9.781062724730324,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 24482
    },
    {
      "epoch": 9.781462245305633,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 24483
    },
    {
      "epoch": 9.781861765880944,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 24484
    },
    {
      "epoch": 9.782261286456253,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 24485
    },
    {
      "epoch": 9.782660807031561,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 24486
    },
    {
      "epoch": 9.783060327606872,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0503,
      "step": 24487
    },
    {
      "epoch": 9.783459848182181,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 24488
    },
    {
      "epoch": 9.783859368757492,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 24489
    },
    {
      "epoch": 9.7842588893328,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 24490
    },
    {
      "epoch": 9.78465840990811,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 24491
    },
    {
      "epoch": 9.78505793048342,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 24492
    },
    {
      "epoch": 9.785457451058729,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 24493
    },
    {
      "epoch": 9.78585697163404,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 24494
    },
    {
      "epoch": 9.786256492209349,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 24495
    },
    {
      "epoch": 9.786656012784658,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 24496
    },
    {
      "epoch": 9.787055533359968,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 24497
    },
    {
      "epoch": 9.787455053935277,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 24498
    },
    {
      "epoch": 9.787854574510588,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 24499
    },
    {
      "epoch": 9.788254095085897,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 24500
    },
    {
      "epoch": 9.788653615661207,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 24501
    },
    {
      "epoch": 9.789053136236516,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 24502
    },
    {
      "epoch": 9.789452656811825,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 24503
    },
    {
      "epoch": 9.789852177387136,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 24504
    },
    {
      "epoch": 9.790251697962445,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 24505
    },
    {
      "epoch": 9.790651218537755,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 24506
    },
    {
      "epoch": 9.791050739113064,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 24507
    },
    {
      "epoch": 9.791450259688373,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 24508
    },
    {
      "epoch": 9.791849780263684,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 24509
    },
    {
      "epoch": 9.792249300838993,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 24510
    },
    {
      "epoch": 9.792648821414303,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 24511
    },
    {
      "epoch": 9.793048341989612,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 24512
    },
    {
      "epoch": 9.793447862564921,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 24513
    },
    {
      "epoch": 9.793847383140232,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 24514
    },
    {
      "epoch": 9.79424690371554,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 24515
    },
    {
      "epoch": 9.794646424290852,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 24516
    },
    {
      "epoch": 9.79504594486616,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 24517
    },
    {
      "epoch": 9.795445465441471,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 24518
    },
    {
      "epoch": 9.79584498601678,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 24519
    },
    {
      "epoch": 9.796244506592089,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 24520
    },
    {
      "epoch": 9.7966440271674,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 24521
    },
    {
      "epoch": 9.797043547742708,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 24522
    },
    {
      "epoch": 9.79744306831802,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 24523
    },
    {
      "epoch": 9.797842588893328,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 24524
    },
    {
      "epoch": 9.798242109468637,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 24525
    },
    {
      "epoch": 9.798641630043948,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 24526
    },
    {
      "epoch": 9.799041150619257,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 24527
    },
    {
      "epoch": 9.799440671194567,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 24528
    },
    {
      "epoch": 9.799840191769876,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 24529
    },
    {
      "epoch": 9.800239712345185,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 24530
    },
    {
      "epoch": 9.800639232920496,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 24531
    },
    {
      "epoch": 9.801038753495805,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 24532
    },
    {
      "epoch": 9.801438274071115,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 24533
    },
    {
      "epoch": 9.801837794646424,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 24534
    },
    {
      "epoch": 9.802237315221735,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 24535
    },
    {
      "epoch": 9.802636835797044,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 24536
    },
    {
      "epoch": 9.803036356372353,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 24537
    },
    {
      "epoch": 9.803435876947663,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 24538
    },
    {
      "epoch": 9.803835397522972,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 24539
    },
    {
      "epoch": 9.804234918098283,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 24540
    },
    {
      "epoch": 9.804634438673592,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 24541
    },
    {
      "epoch": 9.8050339592489,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 24542
    },
    {
      "epoch": 9.805433479824211,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 24543
    },
    {
      "epoch": 9.80583300039952,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 24544
    },
    {
      "epoch": 9.806232520974831,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 24545
    },
    {
      "epoch": 9.80663204155014,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 24546
    },
    {
      "epoch": 9.807031562125449,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 24547
    },
    {
      "epoch": 9.80743108270076,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 24548
    },
    {
      "epoch": 9.807830603276068,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 24549
    },
    {
      "epoch": 9.808230123851379,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 24550
    },
    {
      "epoch": 9.808629644426688,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 24551
    },
    {
      "epoch": 9.809029165001999,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 24552
    },
    {
      "epoch": 9.809428685577307,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 24553
    },
    {
      "epoch": 9.809828206152616,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 24554
    },
    {
      "epoch": 9.810227726727927,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 24555
    },
    {
      "epoch": 9.810627247303236,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 24556
    },
    {
      "epoch": 9.811026767878547,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 24557
    },
    {
      "epoch": 9.811426288453855,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 24558
    },
    {
      "epoch": 9.811825809029164,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 24559
    },
    {
      "epoch": 9.812225329604475,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 24560
    },
    {
      "epoch": 9.812624850179784,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 24561
    },
    {
      "epoch": 9.813024370755095,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 24562
    },
    {
      "epoch": 9.813423891330403,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 24563
    },
    {
      "epoch": 9.813823411905712,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 24564
    },
    {
      "epoch": 9.814222932481023,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 24565
    },
    {
      "epoch": 9.814622453056332,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 24566
    },
    {
      "epoch": 9.815021973631643,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 24567
    },
    {
      "epoch": 9.815421494206952,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 24568
    },
    {
      "epoch": 9.815821014782262,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 24569
    },
    {
      "epoch": 9.816220535357571,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 24570
    },
    {
      "epoch": 9.81662005593288,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 24571
    },
    {
      "epoch": 9.81701957650819,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 24572
    },
    {
      "epoch": 9.8174190970835,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 24573
    },
    {
      "epoch": 9.81781861765881,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 24574
    },
    {
      "epoch": 9.81821813823412,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 24575
    },
    {
      "epoch": 9.818617658809428,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 24576
    },
    {
      "epoch": 9.819017179384739,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 24577
    },
    {
      "epoch": 9.819416699960048,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 24578
    },
    {
      "epoch": 9.819816220535358,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 24579
    },
    {
      "epoch": 9.820215741110667,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 24580
    },
    {
      "epoch": 9.820615261685976,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 24581
    },
    {
      "epoch": 9.821014782261287,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 24582
    },
    {
      "epoch": 9.821414302836596,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 24583
    },
    {
      "epoch": 9.821813823411906,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 24584
    },
    {
      "epoch": 9.822213343987215,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 24585
    },
    {
      "epoch": 9.822612864562524,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 24586
    },
    {
      "epoch": 9.823012385137835,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 24587
    },
    {
      "epoch": 9.823411905713144,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 24588
    },
    {
      "epoch": 9.823811426288454,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 24589
    },
    {
      "epoch": 9.824210946863763,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 24590
    },
    {
      "epoch": 9.824610467439072,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 24591
    },
    {
      "epoch": 9.825009988014383,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 24592
    },
    {
      "epoch": 9.825409508589692,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 24593
    },
    {
      "epoch": 9.825809029165002,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 24594
    },
    {
      "epoch": 9.826208549740311,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 24595
    },
    {
      "epoch": 9.826608070315622,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 24596
    },
    {
      "epoch": 9.827007590890931,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 24597
    },
    {
      "epoch": 9.82740711146624,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 24598
    },
    {
      "epoch": 9.82780663204155,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 24599
    },
    {
      "epoch": 9.82820615261686,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 24600
    },
    {
      "epoch": 9.82860567319217,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 24601
    },
    {
      "epoch": 9.829005193767479,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 24602
    },
    {
      "epoch": 9.829404714342788,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 24603
    },
    {
      "epoch": 9.829804234918099,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 24604
    },
    {
      "epoch": 9.830203755493407,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 24605
    },
    {
      "epoch": 9.830603276068718,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 24606
    },
    {
      "epoch": 9.831002796644027,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 24607
    },
    {
      "epoch": 9.831402317219336,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 24608
    },
    {
      "epoch": 9.831801837794647,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 24609
    },
    {
      "epoch": 9.832201358369955,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 24610
    },
    {
      "epoch": 9.832600878945266,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 24611
    },
    {
      "epoch": 9.833000399520575,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 24612
    },
    {
      "epoch": 9.833399920095886,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 24613
    },
    {
      "epoch": 9.833799440671195,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 24614
    },
    {
      "epoch": 9.834198961246504,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 24615
    },
    {
      "epoch": 9.834598481821814,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 24616
    },
    {
      "epoch": 9.834998002397123,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 24617
    },
    {
      "epoch": 9.835397522972434,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 24618
    },
    {
      "epoch": 9.835797043547743,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 24619
    },
    {
      "epoch": 9.836196564123052,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 24620
    },
    {
      "epoch": 9.836596084698362,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 24621
    },
    {
      "epoch": 9.836995605273671,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 24622
    },
    {
      "epoch": 9.837395125848982,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 24623
    },
    {
      "epoch": 9.83779464642429,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 24624
    },
    {
      "epoch": 9.8381941669996,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 24625
    },
    {
      "epoch": 9.83859368757491,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 24626
    },
    {
      "epoch": 9.83899320815022,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 24627
    },
    {
      "epoch": 9.83939272872553,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 24628
    },
    {
      "epoch": 9.839792249300839,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 24629
    },
    {
      "epoch": 9.84019176987615,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 24630
    },
    {
      "epoch": 9.840591290451458,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 24631
    },
    {
      "epoch": 9.840990811026767,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 24632
    },
    {
      "epoch": 9.841390331602078,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 24633
    },
    {
      "epoch": 9.841789852177387,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 24634
    },
    {
      "epoch": 9.842189372752697,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 24635
    },
    {
      "epoch": 9.842588893328006,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 24636
    },
    {
      "epoch": 9.842988413903315,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 24637
    },
    {
      "epoch": 9.843387934478626,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 24638
    },
    {
      "epoch": 9.843787455053935,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 24639
    },
    {
      "epoch": 9.844186975629245,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 24640
    },
    {
      "epoch": 9.844586496204554,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 24641
    },
    {
      "epoch": 9.844986016779863,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 24642
    },
    {
      "epoch": 9.845385537355174,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 24643
    },
    {
      "epoch": 9.845785057930483,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 24644
    },
    {
      "epoch": 9.846184578505794,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 24645
    },
    {
      "epoch": 9.846584099081102,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 24646
    },
    {
      "epoch": 9.846983619656413,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 24647
    },
    {
      "epoch": 9.847383140231722,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 24648
    },
    {
      "epoch": 9.847782660807031,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 24649
    },
    {
      "epoch": 9.848182181382342,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 24650
    },
    {
      "epoch": 9.84858170195765,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 24651
    },
    {
      "epoch": 9.848981222532961,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 24652
    },
    {
      "epoch": 9.84938074310827,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 24653
    },
    {
      "epoch": 9.849780263683579,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 24654
    },
    {
      "epoch": 9.85017978425889,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 24655
    },
    {
      "epoch": 9.850579304834199,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 24656
    },
    {
      "epoch": 9.85097882540951,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 24657
    },
    {
      "epoch": 9.851378345984818,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 24658
    },
    {
      "epoch": 9.851777866560127,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 24659
    },
    {
      "epoch": 9.852177387135438,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 24660
    },
    {
      "epoch": 9.852576907710747,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 24661
    },
    {
      "epoch": 9.852976428286057,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 24662
    },
    {
      "epoch": 9.853375948861366,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 24663
    },
    {
      "epoch": 9.853775469436677,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 24664
    },
    {
      "epoch": 9.854174990011986,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 24665
    },
    {
      "epoch": 9.854574510587295,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 24666
    },
    {
      "epoch": 9.854974031162605,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 24667
    },
    {
      "epoch": 9.855373551737914,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 24668
    },
    {
      "epoch": 9.855773072313225,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 24669
    },
    {
      "epoch": 9.856172592888534,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 24670
    },
    {
      "epoch": 9.856572113463843,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 24671
    },
    {
      "epoch": 9.856971634039153,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 24672
    },
    {
      "epoch": 9.857371154614462,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 24673
    },
    {
      "epoch": 9.857770675189773,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 24674
    },
    {
      "epoch": 9.858170195765082,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 24675
    },
    {
      "epoch": 9.85856971634039,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 24676
    },
    {
      "epoch": 9.858969236915701,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 24677
    },
    {
      "epoch": 9.85936875749101,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 24678
    },
    {
      "epoch": 9.859768278066321,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 24679
    },
    {
      "epoch": 9.86016779864163,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 24680
    },
    {
      "epoch": 9.86056731921694,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 24681
    },
    {
      "epoch": 9.86096683979225,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 24682
    },
    {
      "epoch": 9.861366360367558,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 24683
    },
    {
      "epoch": 9.861765880942869,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 24684
    },
    {
      "epoch": 9.862165401518178,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 24685
    },
    {
      "epoch": 9.862564922093489,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 24686
    },
    {
      "epoch": 9.862964442668797,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 24687
    },
    {
      "epoch": 9.863363963244106,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 24688
    },
    {
      "epoch": 9.863763483819417,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 24689
    },
    {
      "epoch": 9.864163004394726,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 24690
    },
    {
      "epoch": 9.864562524970037,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 24691
    },
    {
      "epoch": 9.864962045545346,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 24692
    },
    {
      "epoch": 9.865361566120654,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 24693
    },
    {
      "epoch": 9.865761086695965,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 24694
    },
    {
      "epoch": 9.866160607271274,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 24695
    },
    {
      "epoch": 9.866560127846585,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 24696
    },
    {
      "epoch": 9.866959648421894,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 24697
    },
    {
      "epoch": 9.867359168997204,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 24698
    },
    {
      "epoch": 9.867758689572513,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 24699
    },
    {
      "epoch": 9.868158210147822,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 24700
    },
    {
      "epoch": 9.868557730723133,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 24701
    },
    {
      "epoch": 9.868957251298442,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 24702
    },
    {
      "epoch": 9.869356771873752,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 24703
    },
    {
      "epoch": 9.869756292449061,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 24704
    },
    {
      "epoch": 9.87015581302437,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 24705
    },
    {
      "epoch": 9.87055533359968,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 24706
    },
    {
      "epoch": 9.87095485417499,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 24707
    },
    {
      "epoch": 9.8713543747503,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 24708
    },
    {
      "epoch": 9.87175389532561,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 24709
    },
    {
      "epoch": 9.872153415900918,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 24710
    },
    {
      "epoch": 9.872552936476229,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 24711
    },
    {
      "epoch": 9.872952457051538,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 24712
    },
    {
      "epoch": 9.873351977626848,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 24713
    },
    {
      "epoch": 9.873751498202157,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 24714
    },
    {
      "epoch": 9.874151018777468,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 24715
    },
    {
      "epoch": 9.874550539352777,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 24716
    },
    {
      "epoch": 9.874950059928086,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 24717
    },
    {
      "epoch": 9.875349580503396,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 24718
    },
    {
      "epoch": 9.875749101078705,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 24719
    },
    {
      "epoch": 9.876148621654016,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 24720
    },
    {
      "epoch": 9.876548142229325,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 24721
    },
    {
      "epoch": 9.876947662804634,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 24722
    },
    {
      "epoch": 9.877347183379944,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 24723
    },
    {
      "epoch": 9.877746703955253,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 24724
    },
    {
      "epoch": 9.878146224530564,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 24725
    },
    {
      "epoch": 9.878545745105873,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 24726
    },
    {
      "epoch": 9.878945265681182,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 24727
    },
    {
      "epoch": 9.879344786256492,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 24728
    },
    {
      "epoch": 9.879744306831801,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 24729
    },
    {
      "epoch": 9.880143827407112,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 24730
    },
    {
      "epoch": 9.880543347982421,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 24731
    },
    {
      "epoch": 9.880942868557732,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 24732
    },
    {
      "epoch": 9.88134238913304,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 24733
    },
    {
      "epoch": 9.88174190970835,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 24734
    },
    {
      "epoch": 9.88214143028366,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 24735
    },
    {
      "epoch": 9.882540950858969,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 24736
    },
    {
      "epoch": 9.88294047143428,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 24737
    },
    {
      "epoch": 9.883339992009589,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 24738
    },
    {
      "epoch": 9.883739512584897,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 24739
    },
    {
      "epoch": 9.884139033160208,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 24740
    },
    {
      "epoch": 9.884538553735517,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 24741
    },
    {
      "epoch": 9.884938074310828,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 24742
    },
    {
      "epoch": 9.885337594886137,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 24743
    },
    {
      "epoch": 9.885737115461446,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 24744
    },
    {
      "epoch": 9.886136636036756,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 24745
    },
    {
      "epoch": 9.886536156612065,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 24746
    },
    {
      "epoch": 9.886935677187376,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 24747
    },
    {
      "epoch": 9.887335197762685,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0135,
      "step": 24748
    },
    {
      "epoch": 9.887734718337995,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 24749
    },
    {
      "epoch": 9.888134238913304,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 24750
    },
    {
      "epoch": 9.888533759488613,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 24751
    },
    {
      "epoch": 9.888933280063924,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 24752
    },
    {
      "epoch": 9.889332800639233,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 24753
    },
    {
      "epoch": 9.889732321214543,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 24754
    },
    {
      "epoch": 9.890131841789852,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 24755
    },
    {
      "epoch": 9.890531362365161,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 24756
    },
    {
      "epoch": 9.890930882940472,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 24757
    },
    {
      "epoch": 9.89133040351578,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 24758
    },
    {
      "epoch": 9.891729924091091,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 24759
    },
    {
      "epoch": 9.8921294446664,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 24760
    },
    {
      "epoch": 9.89252896524171,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 24761
    },
    {
      "epoch": 9.89292848581702,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 24762
    },
    {
      "epoch": 9.893328006392329,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 24763
    },
    {
      "epoch": 9.89372752696764,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 24764
    },
    {
      "epoch": 9.894127047542948,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 24765
    },
    {
      "epoch": 9.894526568118257,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 24766
    },
    {
      "epoch": 9.894926088693568,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 24767
    },
    {
      "epoch": 9.895325609268877,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 24768
    },
    {
      "epoch": 9.895725129844188,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 24769
    },
    {
      "epoch": 9.896124650419496,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 24770
    },
    {
      "epoch": 9.896524170994805,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 24771
    },
    {
      "epoch": 9.896923691570116,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 24772
    },
    {
      "epoch": 9.897323212145425,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 24773
    },
    {
      "epoch": 9.897722732720736,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 24774
    },
    {
      "epoch": 9.898122253296044,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 24775
    },
    {
      "epoch": 9.898521773871355,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 24776
    },
    {
      "epoch": 9.898921294446664,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 24777
    },
    {
      "epoch": 9.899320815021973,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 24778
    },
    {
      "epoch": 9.899720335597284,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 24779
    },
    {
      "epoch": 9.900119856172592,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 24780
    },
    {
      "epoch": 9.900519376747903,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 24781
    },
    {
      "epoch": 9.900918897323212,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 24782
    },
    {
      "epoch": 9.901318417898521,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 24783
    },
    {
      "epoch": 9.901717938473832,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 24784
    },
    {
      "epoch": 9.90211745904914,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 24785
    },
    {
      "epoch": 9.902516979624451,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 24786
    },
    {
      "epoch": 9.90291650019976,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 24787
    },
    {
      "epoch": 9.903316020775069,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 24788
    },
    {
      "epoch": 9.90371554135038,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 24789
    },
    {
      "epoch": 9.904115061925689,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 24790
    },
    {
      "epoch": 9.904514582501,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 24791
    },
    {
      "epoch": 9.904914103076308,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 24792
    },
    {
      "epoch": 9.905313623651619,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 24793
    },
    {
      "epoch": 9.905713144226928,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 24794
    },
    {
      "epoch": 9.906112664802237,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 24795
    },
    {
      "epoch": 9.906512185377547,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 24796
    },
    {
      "epoch": 9.906911705952856,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 24797
    },
    {
      "epoch": 9.907311226528167,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 24798
    },
    {
      "epoch": 9.907710747103476,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 24799
    },
    {
      "epoch": 9.908110267678785,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 24800
    },
    {
      "epoch": 9.908509788254095,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 24801
    },
    {
      "epoch": 9.908909308829404,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 24802
    },
    {
      "epoch": 9.909308829404715,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 24803
    },
    {
      "epoch": 9.909708349980024,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 24804
    },
    {
      "epoch": 9.910107870555333,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 24805
    },
    {
      "epoch": 9.910507391130643,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 24806
    },
    {
      "epoch": 9.910906911705952,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 24807
    },
    {
      "epoch": 9.911306432281263,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 24808
    },
    {
      "epoch": 9.911705952856572,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 24809
    },
    {
      "epoch": 9.912105473431883,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 24810
    },
    {
      "epoch": 9.912504994007191,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 24811
    },
    {
      "epoch": 9.9129045145825,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 24812
    },
    {
      "epoch": 9.913304035157811,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 24813
    },
    {
      "epoch": 9.91370355573312,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 24814
    },
    {
      "epoch": 9.91410307630843,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 24815
    },
    {
      "epoch": 9.91450259688374,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 24816
    },
    {
      "epoch": 9.914902117459048,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 24817
    },
    {
      "epoch": 9.915301638034359,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 24818
    },
    {
      "epoch": 9.915701158609668,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 24819
    },
    {
      "epoch": 9.916100679184979,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 24820
    },
    {
      "epoch": 9.916500199760288,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 24821
    },
    {
      "epoch": 9.916899720335596,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 24822
    },
    {
      "epoch": 9.917299240910907,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 24823
    },
    {
      "epoch": 9.917698761486216,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 24824
    },
    {
      "epoch": 9.918098282061527,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 24825
    },
    {
      "epoch": 9.918497802636836,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 24826
    },
    {
      "epoch": 9.918897323212146,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 24827
    },
    {
      "epoch": 9.919296843787455,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 24828
    },
    {
      "epoch": 9.919696364362764,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 24829
    },
    {
      "epoch": 9.920095884938075,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 24830
    },
    {
      "epoch": 9.920495405513384,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 24831
    },
    {
      "epoch": 9.920894926088694,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 24832
    },
    {
      "epoch": 9.921294446664003,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 24833
    },
    {
      "epoch": 9.921693967239312,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 24834
    },
    {
      "epoch": 9.922093487814623,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 24835
    },
    {
      "epoch": 9.922493008389932,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 24836
    },
    {
      "epoch": 9.922892528965242,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 24837
    },
    {
      "epoch": 9.923292049540551,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 24838
    },
    {
      "epoch": 9.92369157011586,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 24839
    },
    {
      "epoch": 9.92409109069117,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 24840
    },
    {
      "epoch": 9.92449061126648,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 24841
    },
    {
      "epoch": 9.92489013184179,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 24842
    },
    {
      "epoch": 9.9252896524171,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 24843
    },
    {
      "epoch": 9.92568917299241,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 24844
    },
    {
      "epoch": 9.926088693567719,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 24845
    },
    {
      "epoch": 9.926488214143028,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 24846
    },
    {
      "epoch": 9.926887734718338,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 24847
    },
    {
      "epoch": 9.927287255293647,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 24848
    },
    {
      "epoch": 9.927686775868958,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 24849
    },
    {
      "epoch": 9.928086296444267,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 24850
    },
    {
      "epoch": 9.928485817019576,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 24851
    },
    {
      "epoch": 9.928885337594886,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 24852
    },
    {
      "epoch": 9.929284858170195,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 24853
    },
    {
      "epoch": 9.929684378745506,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 24854
    },
    {
      "epoch": 9.930083899320815,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 24855
    },
    {
      "epoch": 9.930483419896124,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 24856
    },
    {
      "epoch": 9.930882940471434,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 24857
    },
    {
      "epoch": 9.931282461046743,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 24858
    },
    {
      "epoch": 9.931681981622054,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 24859
    },
    {
      "epoch": 9.932081502197363,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 24860
    },
    {
      "epoch": 9.932481022772674,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 24861
    },
    {
      "epoch": 9.932880543347983,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 24862
    },
    {
      "epoch": 9.933280063923291,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 24863
    },
    {
      "epoch": 9.933679584498602,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 24864
    },
    {
      "epoch": 9.934079105073911,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 24865
    },
    {
      "epoch": 9.934478625649222,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 24866
    },
    {
      "epoch": 9.93487814622453,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 24867
    },
    {
      "epoch": 9.93527766679984,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 24868
    },
    {
      "epoch": 9.93567718737515,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 24869
    },
    {
      "epoch": 9.936076707950459,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 24870
    },
    {
      "epoch": 9.93647622852577,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 24871
    },
    {
      "epoch": 9.936875749101079,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 24872
    },
    {
      "epoch": 9.937275269676388,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 24873
    },
    {
      "epoch": 9.937674790251698,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 24874
    },
    {
      "epoch": 9.938074310827007,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 24875
    },
    {
      "epoch": 9.938473831402318,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 24876
    },
    {
      "epoch": 9.938873351977627,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 24877
    },
    {
      "epoch": 9.939272872552937,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 24878
    },
    {
      "epoch": 9.939672393128246,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 24879
    },
    {
      "epoch": 9.940071913703555,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 24880
    },
    {
      "epoch": 9.940471434278866,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 24881
    },
    {
      "epoch": 9.940870954854175,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 24882
    },
    {
      "epoch": 9.941270475429485,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 24883
    },
    {
      "epoch": 9.941669996004794,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 24884
    },
    {
      "epoch": 9.942069516580103,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 24885
    },
    {
      "epoch": 9.942469037155414,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 24886
    },
    {
      "epoch": 9.942868557730723,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 24887
    },
    {
      "epoch": 9.943268078306033,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 24888
    },
    {
      "epoch": 9.943667598881342,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 24889
    },
    {
      "epoch": 9.944067119456651,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 24890
    },
    {
      "epoch": 9.944466640031962,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 24891
    },
    {
      "epoch": 9.94486616060727,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 24892
    },
    {
      "epoch": 9.945265681182581,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 24893
    },
    {
      "epoch": 9.94566520175789,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 24894
    },
    {
      "epoch": 9.946064722333201,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 24895
    },
    {
      "epoch": 9.94646424290851,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 24896
    },
    {
      "epoch": 9.946863763483819,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 24897
    },
    {
      "epoch": 9.94726328405913,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 24898
    },
    {
      "epoch": 9.947662804634438,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 24899
    },
    {
      "epoch": 9.948062325209749,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 24900
    },
    {
      "epoch": 9.948461845785058,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 24901
    },
    {
      "epoch": 9.948861366360367,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 24902
    },
    {
      "epoch": 9.949260886935678,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 24903
    },
    {
      "epoch": 9.949660407510986,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 24904
    },
    {
      "epoch": 9.950059928086297,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 24905
    },
    {
      "epoch": 9.950459448661606,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 24906
    },
    {
      "epoch": 9.950858969236915,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 24907
    },
    {
      "epoch": 9.951258489812226,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 24908
    },
    {
      "epoch": 9.951658010387535,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 24909
    },
    {
      "epoch": 9.952057530962845,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 24910
    },
    {
      "epoch": 9.952457051538154,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 24911
    },
    {
      "epoch": 9.952856572113465,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 24912
    },
    {
      "epoch": 9.953256092688774,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 24913
    },
    {
      "epoch": 9.953655613264083,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 24914
    },
    {
      "epoch": 9.954055133839393,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 24915
    },
    {
      "epoch": 9.954454654414702,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 24916
    },
    {
      "epoch": 9.954854174990013,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 24917
    },
    {
      "epoch": 9.955253695565322,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 24918
    },
    {
      "epoch": 9.95565321614063,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 24919
    },
    {
      "epoch": 9.956052736715941,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 24920
    },
    {
      "epoch": 9.95645225729125,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 24921
    },
    {
      "epoch": 9.95685177786656,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 24922
    },
    {
      "epoch": 9.95725129844187,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 24923
    },
    {
      "epoch": 9.957650819017179,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 24924
    },
    {
      "epoch": 9.95805033959249,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 24925
    },
    {
      "epoch": 9.958449860167798,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 24926
    },
    {
      "epoch": 9.958849380743109,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 24927
    },
    {
      "epoch": 9.959248901318418,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 24928
    },
    {
      "epoch": 9.959648421893728,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 24929
    },
    {
      "epoch": 9.960047942469037,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 24930
    },
    {
      "epoch": 9.960447463044346,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 24931
    },
    {
      "epoch": 9.960846983619657,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 24932
    },
    {
      "epoch": 9.961246504194966,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 24933
    },
    {
      "epoch": 9.961646024770276,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 24934
    },
    {
      "epoch": 9.962045545345585,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 24935
    },
    {
      "epoch": 9.962445065920894,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 24936
    },
    {
      "epoch": 9.962844586496205,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 24937
    },
    {
      "epoch": 9.963244107071514,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 24938
    },
    {
      "epoch": 9.963643627646825,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 24939
    },
    {
      "epoch": 9.964043148222133,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 24940
    },
    {
      "epoch": 9.964442668797442,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 24941
    },
    {
      "epoch": 9.964842189372753,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 24942
    },
    {
      "epoch": 9.965241709948062,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 24943
    },
    {
      "epoch": 9.965641230523373,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 24944
    },
    {
      "epoch": 9.966040751098681,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 24945
    },
    {
      "epoch": 9.96644027167399,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 24946
    },
    {
      "epoch": 9.966839792249301,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 24947
    },
    {
      "epoch": 9.96723931282461,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 24948
    },
    {
      "epoch": 9.96763883339992,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 24949
    },
    {
      "epoch": 9.96803835397523,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 24950
    },
    {
      "epoch": 9.968437874550538,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 24951
    },
    {
      "epoch": 9.968837395125849,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 24952
    },
    {
      "epoch": 9.969236915701158,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 24953
    },
    {
      "epoch": 9.969636436276469,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 24954
    },
    {
      "epoch": 9.970035956851778,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 24955
    },
    {
      "epoch": 9.970435477427088,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 24956
    },
    {
      "epoch": 9.970834998002397,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 24957
    },
    {
      "epoch": 9.971234518577706,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 24958
    },
    {
      "epoch": 9.971634039153017,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 24959
    },
    {
      "epoch": 9.972033559728326,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 24960
    },
    {
      "epoch": 9.972433080303636,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 24961
    },
    {
      "epoch": 9.972832600878945,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 24962
    },
    {
      "epoch": 9.973232121454254,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 24963
    },
    {
      "epoch": 9.973631642029565,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 24964
    },
    {
      "epoch": 9.974031162604874,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 24965
    },
    {
      "epoch": 9.974430683180184,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 24966
    },
    {
      "epoch": 9.974830203755493,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 24967
    },
    {
      "epoch": 9.975229724330802,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 24968
    },
    {
      "epoch": 9.975629244906113,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 24969
    },
    {
      "epoch": 9.976028765481422,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 24970
    },
    {
      "epoch": 9.976428286056732,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 24971
    },
    {
      "epoch": 9.976827806632041,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 24972
    },
    {
      "epoch": 9.977227327207352,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 24973
    },
    {
      "epoch": 9.97762684778266,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 24974
    },
    {
      "epoch": 9.97802636835797,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 24975
    },
    {
      "epoch": 9.97842588893328,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 24976
    },
    {
      "epoch": 9.97882540950859,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 24977
    },
    {
      "epoch": 9.9792249300839,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 24978
    },
    {
      "epoch": 9.979624450659209,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 24979
    },
    {
      "epoch": 9.980023971234518,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 24980
    },
    {
      "epoch": 9.980423491809828,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 24981
    },
    {
      "epoch": 9.980823012385137,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 24982
    },
    {
      "epoch": 9.981222532960448,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 24983
    },
    {
      "epoch": 9.981622053535757,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 24984
    },
    {
      "epoch": 9.982021574111066,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 24985
    },
    {
      "epoch": 9.982421094686377,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 24986
    },
    {
      "epoch": 9.982820615261685,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 24987
    },
    {
      "epoch": 9.983220135836996,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 24988
    },
    {
      "epoch": 9.983619656412305,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 24989
    },
    {
      "epoch": 9.984019176987616,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 24990
    },
    {
      "epoch": 9.984418697562925,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 24991
    },
    {
      "epoch": 9.984818218138233,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 24992
    },
    {
      "epoch": 9.985217738713544,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 24993
    },
    {
      "epoch": 9.985617259288853,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 24994
    },
    {
      "epoch": 9.986016779864164,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 24995
    },
    {
      "epoch": 9.986416300439473,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 24996
    },
    {
      "epoch": 9.986815821014781,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 24997
    },
    {
      "epoch": 9.987215341590092,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 24998
    },
    {
      "epoch": 9.987614862165401,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 24999
    },
    {
      "epoch": 9.988014382740712,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 25000
    },
    {
      "epoch": 9.98841390331602,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 25001
    },
    {
      "epoch": 9.98881342389133,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 25002
    },
    {
      "epoch": 9.98921294446664,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 25003
    },
    {
      "epoch": 9.989612465041949,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 25004
    },
    {
      "epoch": 9.99001198561726,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 25005
    },
    {
      "epoch": 9.990411506192569,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 25006
    },
    {
      "epoch": 9.99081102676788,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 25007
    },
    {
      "epoch": 9.991210547343188,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 25008
    },
    {
      "epoch": 9.991610067918497,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 25009
    },
    {
      "epoch": 9.992009588493808,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 25010
    },
    {
      "epoch": 9.992409109069117,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 25011
    },
    {
      "epoch": 9.992808629644427,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 25012
    },
    {
      "epoch": 9.993208150219736,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 25013
    },
    {
      "epoch": 9.993607670795045,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 25014
    },
    {
      "epoch": 9.994007191370356,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 25015
    },
    {
      "epoch": 9.994406711945665,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 25016
    },
    {
      "epoch": 9.994806232520975,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 25017
    },
    {
      "epoch": 9.995205753096284,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 25018
    },
    {
      "epoch": 9.995605273671593,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 25019
    },
    {
      "epoch": 9.996004794246904,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 25020
    },
    {
      "epoch": 9.996404314822213,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 25021
    },
    {
      "epoch": 9.996803835397523,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 25022
    },
    {
      "epoch": 9.997203355972832,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 25023
    },
    {
      "epoch": 9.997602876548143,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 25024
    },
    {
      "epoch": 9.998002397123452,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 25025
    },
    {
      "epoch": 9.99840191769876,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 25026
    },
    {
      "epoch": 9.998801438274072,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 25027
    },
    {
      "epoch": 9.99920095884938,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 25028
    },
    {
      "epoch": 9.999600479424691,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 25029
    },
    {
      "epoch": 10.0,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 25030
    },
    {
      "epoch": 10.000399520575309,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 25031
    },
    {
      "epoch": 10.00079904115062,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 25032
    },
    {
      "epoch": 10.001198561725928,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 25033
    },
    {
      "epoch": 10.00159808230124,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 25034
    },
    {
      "epoch": 10.001997602876548,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 25035
    },
    {
      "epoch": 10.002397123451857,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 25036
    },
    {
      "epoch": 10.002796644027168,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 25037
    },
    {
      "epoch": 10.003196164602477,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 25038
    },
    {
      "epoch": 10.003595685177787,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 25039
    },
    {
      "epoch": 10.003995205753096,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 25040
    },
    {
      "epoch": 10.004394726328407,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 25041
    },
    {
      "epoch": 10.004794246903716,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 25042
    },
    {
      "epoch": 10.005193767479025,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 25043
    },
    {
      "epoch": 10.005593288054335,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 25044
    },
    {
      "epoch": 10.005992808629644,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 25045
    },
    {
      "epoch": 10.006392329204955,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 25046
    },
    {
      "epoch": 10.006791849780264,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 25047
    },
    {
      "epoch": 10.007191370355573,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 25048
    },
    {
      "epoch": 10.007590890930883,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 25049
    },
    {
      "epoch": 10.007990411506192,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 25050
    },
    {
      "epoch": 10.008389932081503,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 25051
    },
    {
      "epoch": 10.008789452656812,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 25052
    },
    {
      "epoch": 10.00918897323212,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 25053
    },
    {
      "epoch": 10.009588493807431,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 25054
    },
    {
      "epoch": 10.00998801438274,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 25055
    },
    {
      "epoch": 10.010387534958051,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 25056
    },
    {
      "epoch": 10.01078705553336,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 25057
    },
    {
      "epoch": 10.01118657610867,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 25058
    },
    {
      "epoch": 10.01158609668398,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 25059
    },
    {
      "epoch": 10.011985617259288,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 25060
    },
    {
      "epoch": 10.012385137834599,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 25061
    },
    {
      "epoch": 10.012784658409908,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 25062
    },
    {
      "epoch": 10.013184178985219,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 25063
    },
    {
      "epoch": 10.013583699560527,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 25064
    },
    {
      "epoch": 10.013983220135836,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 25065
    },
    {
      "epoch": 10.014382740711147,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 25066
    },
    {
      "epoch": 10.014782261286456,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 25067
    },
    {
      "epoch": 10.015181781861767,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 25068
    },
    {
      "epoch": 10.015581302437075,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 25069
    },
    {
      "epoch": 10.015980823012384,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 25070
    },
    {
      "epoch": 10.016380343587695,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 25071
    },
    {
      "epoch": 10.016779864163004,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 25072
    },
    {
      "epoch": 10.017179384738315,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 25073
    },
    {
      "epoch": 10.017578905313623,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 25074
    },
    {
      "epoch": 10.017978425888934,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 25075
    },
    {
      "epoch": 10.018377946464243,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 25076
    },
    {
      "epoch": 10.018777467039552,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 25077
    },
    {
      "epoch": 10.019176987614863,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 25078
    },
    {
      "epoch": 10.019576508190172,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 25079
    },
    {
      "epoch": 10.019976028765482,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 25080
    },
    {
      "epoch": 10.020375549340791,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 25081
    },
    {
      "epoch": 10.0207750699161,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 25082
    },
    {
      "epoch": 10.02117459049141,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 25083
    },
    {
      "epoch": 10.02157411106672,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 25084
    },
    {
      "epoch": 10.02197363164203,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 25085
    },
    {
      "epoch": 10.02237315221734,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 25086
    },
    {
      "epoch": 10.022772672792648,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 25087
    },
    {
      "epoch": 10.023172193367959,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 25088
    },
    {
      "epoch": 10.023571713943268,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0482,
      "step": 25089
    },
    {
      "epoch": 10.023971234518578,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 25090
    },
    {
      "epoch": 10.024370755093887,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 25091
    },
    {
      "epoch": 10.024770275669196,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 25092
    },
    {
      "epoch": 10.025169796244507,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 25093
    },
    {
      "epoch": 10.025569316819816,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 25094
    },
    {
      "epoch": 10.025968837395126,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 25095
    },
    {
      "epoch": 10.026368357970435,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 25096
    },
    {
      "epoch": 10.026767878545746,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 25097
    },
    {
      "epoch": 10.027167399121055,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 25098
    },
    {
      "epoch": 10.027566919696364,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 25099
    },
    {
      "epoch": 10.027966440271674,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 25100
    },
    {
      "epoch": 10.028365960846983,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 25101
    },
    {
      "epoch": 10.028765481422294,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 25102
    },
    {
      "epoch": 10.029165001997603,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 25103
    },
    {
      "epoch": 10.029564522572912,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 25104
    },
    {
      "epoch": 10.029964043148222,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 25105
    },
    {
      "epoch": 10.030363563723531,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 25106
    },
    {
      "epoch": 10.030763084298842,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 25107
    },
    {
      "epoch": 10.031162604874151,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 25108
    },
    {
      "epoch": 10.03156212544946,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 25109
    },
    {
      "epoch": 10.03196164602477,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 25110
    },
    {
      "epoch": 10.03236116660008,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 25111
    },
    {
      "epoch": 10.03276068717539,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 25112
    },
    {
      "epoch": 10.033160207750699,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 25113
    },
    {
      "epoch": 10.03355972832601,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 25114
    },
    {
      "epoch": 10.033959248901319,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 25115
    },
    {
      "epoch": 10.034358769476627,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 25116
    },
    {
      "epoch": 10.034758290051938,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 25117
    },
    {
      "epoch": 10.035157810627247,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 25118
    },
    {
      "epoch": 10.035557331202558,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 25119
    },
    {
      "epoch": 10.035956851777867,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 25120
    },
    {
      "epoch": 10.036356372353175,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 25121
    },
    {
      "epoch": 10.036755892928486,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 25122
    },
    {
      "epoch": 10.037155413503795,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 25123
    },
    {
      "epoch": 10.037554934079106,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 25124
    },
    {
      "epoch": 10.037954454654415,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 25125
    },
    {
      "epoch": 10.038353975229724,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 25126
    },
    {
      "epoch": 10.038753495805034,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 25127
    },
    {
      "epoch": 10.039153016380343,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 25128
    },
    {
      "epoch": 10.039552536955654,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 25129
    },
    {
      "epoch": 10.039952057530963,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 25130
    },
    {
      "epoch": 10.040351578106273,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 25131
    },
    {
      "epoch": 10.040751098681582,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 25132
    },
    {
      "epoch": 10.041150619256891,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.0507,
      "step": 25133
    },
    {
      "epoch": 10.041550139832202,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 25134
    },
    {
      "epoch": 10.04194966040751,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 25135
    },
    {
      "epoch": 10.042349180982821,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 25136
    },
    {
      "epoch": 10.04274870155813,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 25137
    },
    {
      "epoch": 10.04314822213344,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 25138
    },
    {
      "epoch": 10.04354774270875,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 25139
    },
    {
      "epoch": 10.043947263284059,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 25140
    },
    {
      "epoch": 10.04434678385937,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 25141
    },
    {
      "epoch": 10.044746304434678,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 25142
    },
    {
      "epoch": 10.045145825009987,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 25143
    },
    {
      "epoch": 10.045545345585298,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 25144
    },
    {
      "epoch": 10.045944866160607,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 25145
    },
    {
      "epoch": 10.046344386735917,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 25146
    },
    {
      "epoch": 10.046743907311226,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 25147
    },
    {
      "epoch": 10.047143427886537,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 25148
    },
    {
      "epoch": 10.047542948461846,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 25149
    },
    {
      "epoch": 10.047942469037155,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 25150
    },
    {
      "epoch": 10.048341989612465,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 25151
    },
    {
      "epoch": 10.048741510187774,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 25152
    },
    {
      "epoch": 10.049141030763085,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 25153
    },
    {
      "epoch": 10.049540551338394,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 25154
    },
    {
      "epoch": 10.049940071913703,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 25155
    },
    {
      "epoch": 10.050339592489014,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 25156
    },
    {
      "epoch": 10.050739113064322,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 25157
    },
    {
      "epoch": 10.051138633639633,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 25158
    },
    {
      "epoch": 10.051538154214942,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 25159
    },
    {
      "epoch": 10.051937674790251,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 25160
    },
    {
      "epoch": 10.052337195365562,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 25161
    },
    {
      "epoch": 10.05273671594087,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 25162
    },
    {
      "epoch": 10.053136236516181,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 25163
    },
    {
      "epoch": 10.05353575709149,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 25164
    },
    {
      "epoch": 10.053935277666799,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 25165
    },
    {
      "epoch": 10.05433479824211,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 25166
    },
    {
      "epoch": 10.054734318817419,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 25167
    },
    {
      "epoch": 10.05513383939273,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 25168
    },
    {
      "epoch": 10.055533359968038,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 25169
    },
    {
      "epoch": 10.055932880543349,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0528,
      "step": 25170
    },
    {
      "epoch": 10.056332401118658,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 25171
    },
    {
      "epoch": 10.056731921693967,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 25172
    },
    {
      "epoch": 10.057131442269277,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 25173
    },
    {
      "epoch": 10.057530962844586,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 25174
    },
    {
      "epoch": 10.057930483419897,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 25175
    },
    {
      "epoch": 10.058330003995206,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 25176
    },
    {
      "epoch": 10.058729524570515,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 25177
    },
    {
      "epoch": 10.059129045145825,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 25178
    },
    {
      "epoch": 10.059528565721134,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 25179
    },
    {
      "epoch": 10.059928086296445,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 25180
    },
    {
      "epoch": 10.060327606871754,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 25181
    },
    {
      "epoch": 10.060727127447063,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 25182
    },
    {
      "epoch": 10.061126648022373,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 25183
    },
    {
      "epoch": 10.061526168597682,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 25184
    },
    {
      "epoch": 10.061925689172993,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 25185
    },
    {
      "epoch": 10.062325209748302,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 25186
    },
    {
      "epoch": 10.062724730323612,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 25187
    },
    {
      "epoch": 10.063124250898921,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 25188
    },
    {
      "epoch": 10.06352377147423,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 25189
    },
    {
      "epoch": 10.063923292049541,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 25190
    },
    {
      "epoch": 10.06432281262485,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 25191
    },
    {
      "epoch": 10.06472233320016,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 25192
    },
    {
      "epoch": 10.06512185377547,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 25193
    },
    {
      "epoch": 10.065521374350778,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 25194
    },
    {
      "epoch": 10.065920894926089,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 25195
    },
    {
      "epoch": 10.066320415501398,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 25196
    },
    {
      "epoch": 10.066719936076709,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 25197
    },
    {
      "epoch": 10.067119456652017,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 25198
    },
    {
      "epoch": 10.067518977227326,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 25199
    },
    {
      "epoch": 10.067918497802637,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 25200
    },
    {
      "epoch": 10.068318018377946,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 25201
    },
    {
      "epoch": 10.068717538953257,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 25202
    },
    {
      "epoch": 10.069117059528566,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 25203
    },
    {
      "epoch": 10.069516580103876,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 25204
    },
    {
      "epoch": 10.069916100679185,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 25205
    },
    {
      "epoch": 10.070315621254494,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 25206
    },
    {
      "epoch": 10.070715141829805,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 25207
    },
    {
      "epoch": 10.071114662405114,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 25208
    },
    {
      "epoch": 10.071514182980424,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 25209
    },
    {
      "epoch": 10.071913703555733,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 25210
    },
    {
      "epoch": 10.072313224131042,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 25211
    },
    {
      "epoch": 10.072712744706353,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 25212
    },
    {
      "epoch": 10.073112265281662,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 25213
    },
    {
      "epoch": 10.073511785856972,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 25214
    },
    {
      "epoch": 10.073911306432281,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 25215
    },
    {
      "epoch": 10.07431082700759,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 25216
    },
    {
      "epoch": 10.0747103475829,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 25217
    },
    {
      "epoch": 10.07510986815821,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 25218
    },
    {
      "epoch": 10.07550938873352,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 25219
    },
    {
      "epoch": 10.07590890930883,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 25220
    },
    {
      "epoch": 10.07630842988414,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 25221
    },
    {
      "epoch": 10.076707950459449,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 25222
    },
    {
      "epoch": 10.077107471034758,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 25223
    },
    {
      "epoch": 10.077506991610068,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 25224
    },
    {
      "epoch": 10.077906512185377,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 25225
    },
    {
      "epoch": 10.078306032760688,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 25226
    },
    {
      "epoch": 10.078705553335997,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 25227
    },
    {
      "epoch": 10.079105073911306,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 25228
    },
    {
      "epoch": 10.079504594486616,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 25229
    },
    {
      "epoch": 10.079904115061925,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 25230
    },
    {
      "epoch": 10.080303635637236,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 25231
    },
    {
      "epoch": 10.080703156212545,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 25232
    },
    {
      "epoch": 10.081102676787854,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 25233
    },
    {
      "epoch": 10.081502197363164,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 25234
    },
    {
      "epoch": 10.081901717938473,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 25235
    },
    {
      "epoch": 10.082301238513784,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 25236
    },
    {
      "epoch": 10.082700759089093,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 25237
    },
    {
      "epoch": 10.083100279664404,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 25238
    },
    {
      "epoch": 10.083499800239712,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 25239
    },
    {
      "epoch": 10.083899320815021,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 25240
    },
    {
      "epoch": 10.084298841390332,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 25241
    },
    {
      "epoch": 10.084698361965641,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 25242
    },
    {
      "epoch": 10.085097882540952,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 25243
    },
    {
      "epoch": 10.08549740311626,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 25244
    },
    {
      "epoch": 10.08589692369157,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 25245
    },
    {
      "epoch": 10.08629644426688,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 25246
    },
    {
      "epoch": 10.086695964842189,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 25247
    },
    {
      "epoch": 10.0870954854175,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 25248
    },
    {
      "epoch": 10.087495005992809,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 25249
    },
    {
      "epoch": 10.087894526568117,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 25250
    },
    {
      "epoch": 10.088294047143428,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 25251
    },
    {
      "epoch": 10.088693567718737,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 25252
    },
    {
      "epoch": 10.089093088294048,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 25253
    },
    {
      "epoch": 10.089492608869357,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 25254
    },
    {
      "epoch": 10.089892129444667,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 25255
    },
    {
      "epoch": 10.090291650019976,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 25256
    },
    {
      "epoch": 10.090691170595285,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 25257
    },
    {
      "epoch": 10.091090691170596,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 25258
    },
    {
      "epoch": 10.091490211745905,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 25259
    },
    {
      "epoch": 10.091889732321215,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 25260
    },
    {
      "epoch": 10.092289252896524,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 25261
    },
    {
      "epoch": 10.092688773471833,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 25262
    },
    {
      "epoch": 10.093088294047144,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 25263
    },
    {
      "epoch": 10.093487814622453,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 25264
    },
    {
      "epoch": 10.093887335197763,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 25265
    },
    {
      "epoch": 10.094286855773072,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 25266
    },
    {
      "epoch": 10.094686376348381,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 25267
    },
    {
      "epoch": 10.095085896923692,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 25268
    },
    {
      "epoch": 10.095485417499,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 25269
    },
    {
      "epoch": 10.095884938074311,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 25270
    },
    {
      "epoch": 10.09628445864962,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 25271
    },
    {
      "epoch": 10.09668397922493,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 25272
    },
    {
      "epoch": 10.09708349980024,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 25273
    },
    {
      "epoch": 10.097483020375549,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 25274
    },
    {
      "epoch": 10.09788254095086,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 25275
    },
    {
      "epoch": 10.098282061526168,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 25276
    },
    {
      "epoch": 10.098681582101479,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 25277
    },
    {
      "epoch": 10.099081102676788,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 25278
    },
    {
      "epoch": 10.099480623252097,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 25279
    },
    {
      "epoch": 10.099880143827408,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 25280
    },
    {
      "epoch": 10.100279664402716,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 25281
    },
    {
      "epoch": 10.100679184978027,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 25282
    },
    {
      "epoch": 10.101078705553336,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 25283
    },
    {
      "epoch": 10.101478226128645,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 25284
    },
    {
      "epoch": 10.101877746703956,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 25285
    },
    {
      "epoch": 10.102277267279264,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 25286
    },
    {
      "epoch": 10.102676787854575,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 25287
    },
    {
      "epoch": 10.103076308429884,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 25288
    },
    {
      "epoch": 10.103475829005193,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 25289
    },
    {
      "epoch": 10.103875349580504,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 25290
    },
    {
      "epoch": 10.104274870155812,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 25291
    },
    {
      "epoch": 10.104674390731123,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 25292
    },
    {
      "epoch": 10.105073911306432,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 25293
    },
    {
      "epoch": 10.105473431881743,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 25294
    },
    {
      "epoch": 10.105872952457052,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 25295
    },
    {
      "epoch": 10.10627247303236,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 25296
    },
    {
      "epoch": 10.106671993607671,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 25297
    },
    {
      "epoch": 10.10707151418298,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 25298
    },
    {
      "epoch": 10.10747103475829,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 25299
    },
    {
      "epoch": 10.1078705553336,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 25300
    },
    {
      "epoch": 10.108270075908909,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 25301
    },
    {
      "epoch": 10.10866959648422,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 25302
    },
    {
      "epoch": 10.109069117059528,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 25303
    },
    {
      "epoch": 10.109468637634839,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 25304
    },
    {
      "epoch": 10.109868158210148,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 25305
    },
    {
      "epoch": 10.110267678785457,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 25306
    },
    {
      "epoch": 10.110667199360767,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 25307
    },
    {
      "epoch": 10.111066719936076,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 25308
    },
    {
      "epoch": 10.111466240511387,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 25309
    },
    {
      "epoch": 10.111865761086696,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 25310
    },
    {
      "epoch": 10.112265281662006,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 25311
    },
    {
      "epoch": 10.112664802237315,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 25312
    },
    {
      "epoch": 10.113064322812624,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 25313
    },
    {
      "epoch": 10.113463843387935,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 25314
    },
    {
      "epoch": 10.113863363963244,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 25315
    },
    {
      "epoch": 10.114262884538554,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 25316
    },
    {
      "epoch": 10.114662405113863,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 25317
    },
    {
      "epoch": 10.115061925689172,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 25318
    },
    {
      "epoch": 10.115461446264483,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 25319
    },
    {
      "epoch": 10.115860966839792,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 25320
    },
    {
      "epoch": 10.116260487415103,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 25321
    },
    {
      "epoch": 10.116660007990411,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 25322
    },
    {
      "epoch": 10.11705952856572,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 25323
    },
    {
      "epoch": 10.117459049141031,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 25324
    },
    {
      "epoch": 10.11785856971634,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 25325
    },
    {
      "epoch": 10.11825809029165,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 25326
    },
    {
      "epoch": 10.11865761086696,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 25327
    },
    {
      "epoch": 10.11905713144227,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 25328
    },
    {
      "epoch": 10.119456652017579,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 25329
    },
    {
      "epoch": 10.119856172592888,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 25330
    },
    {
      "epoch": 10.120255693168199,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 25331
    },
    {
      "epoch": 10.120655213743508,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 25332
    },
    {
      "epoch": 10.121054734318818,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 25333
    },
    {
      "epoch": 10.121454254894127,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 25334
    },
    {
      "epoch": 10.121853775469436,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 25335
    },
    {
      "epoch": 10.122253296044747,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 25336
    },
    {
      "epoch": 10.122652816620056,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 25337
    },
    {
      "epoch": 10.123052337195366,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 25338
    },
    {
      "epoch": 10.123451857770675,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 25339
    },
    {
      "epoch": 10.123851378345984,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 25340
    },
    {
      "epoch": 10.124250898921295,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 25341
    },
    {
      "epoch": 10.124650419496604,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 25342
    },
    {
      "epoch": 10.125049940071914,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 25343
    },
    {
      "epoch": 10.125449460647223,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 25344
    },
    {
      "epoch": 10.125848981222532,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 25345
    },
    {
      "epoch": 10.126248501797843,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 25346
    },
    {
      "epoch": 10.126648022373152,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 25347
    },
    {
      "epoch": 10.127047542948462,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 25348
    },
    {
      "epoch": 10.127447063523771,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 25349
    },
    {
      "epoch": 10.127846584099082,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 25350
    },
    {
      "epoch": 10.12824610467439,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 25351
    },
    {
      "epoch": 10.1286456252497,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 25352
    },
    {
      "epoch": 10.12904514582501,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 25353
    },
    {
      "epoch": 10.12944466640032,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 25354
    },
    {
      "epoch": 10.12984418697563,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 25355
    },
    {
      "epoch": 10.130243707550939,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 25356
    },
    {
      "epoch": 10.130643228126248,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 25357
    },
    {
      "epoch": 10.131042748701558,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 25358
    },
    {
      "epoch": 10.131442269276867,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 25359
    },
    {
      "epoch": 10.131841789852178,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 25360
    },
    {
      "epoch": 10.132241310427487,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 25361
    },
    {
      "epoch": 10.132640831002796,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 25362
    },
    {
      "epoch": 10.133040351578106,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 25363
    },
    {
      "epoch": 10.133439872153415,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 25364
    },
    {
      "epoch": 10.133839392728726,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 25365
    },
    {
      "epoch": 10.134238913304035,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 25366
    },
    {
      "epoch": 10.134638433879346,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 25367
    },
    {
      "epoch": 10.135037954454654,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 25368
    },
    {
      "epoch": 10.135437475029963,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 25369
    },
    {
      "epoch": 10.135836995605274,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 25370
    },
    {
      "epoch": 10.136236516180583,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 25371
    },
    {
      "epoch": 10.136636036755894,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 25372
    },
    {
      "epoch": 10.137035557331203,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 25373
    },
    {
      "epoch": 10.137435077906511,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 25374
    },
    {
      "epoch": 10.137834598481822,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 25375
    },
    {
      "epoch": 10.138234119057131,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 25376
    },
    {
      "epoch": 10.138633639632442,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 25377
    },
    {
      "epoch": 10.13903316020775,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 25378
    },
    {
      "epoch": 10.13943268078306,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 25379
    },
    {
      "epoch": 10.13983220135837,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 25380
    },
    {
      "epoch": 10.140231721933679,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 25381
    },
    {
      "epoch": 10.14063124250899,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 25382
    },
    {
      "epoch": 10.141030763084299,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 25383
    },
    {
      "epoch": 10.14143028365961,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 25384
    },
    {
      "epoch": 10.141829804234918,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 25385
    },
    {
      "epoch": 10.142229324810227,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 25386
    },
    {
      "epoch": 10.142628845385538,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 25387
    },
    {
      "epoch": 10.143028365960847,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 25388
    },
    {
      "epoch": 10.143427886536157,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 25389
    },
    {
      "epoch": 10.143827407111466,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 25390
    },
    {
      "epoch": 10.144226927686775,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 25391
    },
    {
      "epoch": 10.144626448262086,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 25392
    },
    {
      "epoch": 10.145025968837395,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 25393
    },
    {
      "epoch": 10.145425489412705,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 25394
    },
    {
      "epoch": 10.145825009988014,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 25395
    },
    {
      "epoch": 10.146224530563323,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 25396
    },
    {
      "epoch": 10.146624051138634,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 25397
    },
    {
      "epoch": 10.147023571713943,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 25398
    },
    {
      "epoch": 10.147423092289253,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 25399
    },
    {
      "epoch": 10.147822612864562,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 25400
    },
    {
      "epoch": 10.148222133439873,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 25401
    },
    {
      "epoch": 10.148621654015182,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 25402
    },
    {
      "epoch": 10.14902117459049,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 25403
    },
    {
      "epoch": 10.149420695165801,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 25404
    },
    {
      "epoch": 10.14982021574111,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 25405
    },
    {
      "epoch": 10.150219736316421,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 25406
    },
    {
      "epoch": 10.15061925689173,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 25407
    },
    {
      "epoch": 10.151018777467039,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 25408
    },
    {
      "epoch": 10.15141829804235,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 25409
    },
    {
      "epoch": 10.151817818617658,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 25410
    },
    {
      "epoch": 10.152217339192969,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 25411
    },
    {
      "epoch": 10.152616859768278,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 25412
    },
    {
      "epoch": 10.153016380343587,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 25413
    },
    {
      "epoch": 10.153415900918898,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 25414
    },
    {
      "epoch": 10.153815421494206,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 25415
    },
    {
      "epoch": 10.154214942069517,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 25416
    },
    {
      "epoch": 10.154614462644826,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 25417
    },
    {
      "epoch": 10.155013983220137,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0522,
      "step": 25418
    },
    {
      "epoch": 10.155413503795446,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 25419
    },
    {
      "epoch": 10.155813024370755,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 25420
    },
    {
      "epoch": 10.156212544946065,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 25421
    },
    {
      "epoch": 10.156612065521374,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 25422
    },
    {
      "epoch": 10.157011586096685,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 25423
    },
    {
      "epoch": 10.157411106671994,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 25424
    },
    {
      "epoch": 10.157810627247303,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 25425
    },
    {
      "epoch": 10.158210147822613,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 25426
    },
    {
      "epoch": 10.158609668397922,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 25427
    },
    {
      "epoch": 10.159009188973233,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 25428
    },
    {
      "epoch": 10.159408709548542,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 25429
    },
    {
      "epoch": 10.15980823012385,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 25430
    },
    {
      "epoch": 10.160207750699161,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 25431
    },
    {
      "epoch": 10.16060727127447,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 25432
    },
    {
      "epoch": 10.16100679184978,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 25433
    },
    {
      "epoch": 10.16140631242509,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 25434
    },
    {
      "epoch": 10.1618058330004,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 25435
    },
    {
      "epoch": 10.16220535357571,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 25436
    },
    {
      "epoch": 10.162604874151018,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 25437
    },
    {
      "epoch": 10.163004394726329,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 25438
    },
    {
      "epoch": 10.163403915301638,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 25439
    },
    {
      "epoch": 10.163803435876948,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 25440
    },
    {
      "epoch": 10.164202956452257,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 25441
    },
    {
      "epoch": 10.164602477027566,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 25442
    },
    {
      "epoch": 10.165001997602877,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 25443
    },
    {
      "epoch": 10.165401518178186,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 25444
    },
    {
      "epoch": 10.165801038753496,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 25445
    },
    {
      "epoch": 10.166200559328805,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 25446
    },
    {
      "epoch": 10.166600079904114,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 25447
    },
    {
      "epoch": 10.166999600479425,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 25448
    },
    {
      "epoch": 10.167399121054734,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 25449
    },
    {
      "epoch": 10.167798641630045,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 25450
    },
    {
      "epoch": 10.168198162205353,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 25451
    },
    {
      "epoch": 10.168597682780662,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 25452
    },
    {
      "epoch": 10.168997203355973,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 25453
    },
    {
      "epoch": 10.169396723931282,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 25454
    },
    {
      "epoch": 10.169796244506593,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 25455
    },
    {
      "epoch": 10.170195765081901,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 25456
    },
    {
      "epoch": 10.170595285657212,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 25457
    },
    {
      "epoch": 10.170994806232521,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 25458
    },
    {
      "epoch": 10.17139432680783,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 25459
    },
    {
      "epoch": 10.17179384738314,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 25460
    },
    {
      "epoch": 10.17219336795845,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 25461
    },
    {
      "epoch": 10.17259288853376,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 25462
    },
    {
      "epoch": 10.172992409109069,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 25463
    },
    {
      "epoch": 10.173391929684378,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 25464
    },
    {
      "epoch": 10.173791450259689,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 25465
    },
    {
      "epoch": 10.174190970834998,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 25466
    },
    {
      "epoch": 10.174590491410308,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 25467
    },
    {
      "epoch": 10.174990011985617,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 25468
    },
    {
      "epoch": 10.175389532560926,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 25469
    },
    {
      "epoch": 10.175789053136237,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 25470
    },
    {
      "epoch": 10.176188573711546,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 25471
    },
    {
      "epoch": 10.176588094286856,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 25472
    },
    {
      "epoch": 10.176987614862165,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 25473
    },
    {
      "epoch": 10.177387135437476,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 25474
    },
    {
      "epoch": 10.177786656012785,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 25475
    },
    {
      "epoch": 10.178186176588094,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 25476
    },
    {
      "epoch": 10.178585697163404,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 25477
    },
    {
      "epoch": 10.178985217738713,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 25478
    },
    {
      "epoch": 10.179384738314024,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 25479
    },
    {
      "epoch": 10.179784258889333,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 25480
    },
    {
      "epoch": 10.180183779464642,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 25481
    },
    {
      "epoch": 10.180583300039952,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 25482
    },
    {
      "epoch": 10.180982820615261,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 25483
    },
    {
      "epoch": 10.181382341190572,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 25484
    },
    {
      "epoch": 10.18178186176588,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 25485
    },
    {
      "epoch": 10.18218138234119,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 25486
    },
    {
      "epoch": 10.1825809029165,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 25487
    },
    {
      "epoch": 10.18298042349181,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 25488
    },
    {
      "epoch": 10.18337994406712,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 25489
    },
    {
      "epoch": 10.183779464642429,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 25490
    },
    {
      "epoch": 10.18417898521774,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 25491
    },
    {
      "epoch": 10.184578505793048,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 25492
    },
    {
      "epoch": 10.184978026368357,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 25493
    },
    {
      "epoch": 10.185377546943668,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 25494
    },
    {
      "epoch": 10.185777067518977,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 25495
    },
    {
      "epoch": 10.186176588094288,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 25496
    },
    {
      "epoch": 10.186576108669597,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 25497
    },
    {
      "epoch": 10.186975629244905,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 25498
    },
    {
      "epoch": 10.187375149820216,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 25499
    },
    {
      "epoch": 10.187774670395525,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 25500
    },
    {
      "epoch": 10.188174190970836,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 25501
    },
    {
      "epoch": 10.188573711546145,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 25502
    },
    {
      "epoch": 10.188973232121453,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 25503
    },
    {
      "epoch": 10.189372752696764,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 25504
    },
    {
      "epoch": 10.189772273272073,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 25505
    },
    {
      "epoch": 10.190171793847384,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 25506
    },
    {
      "epoch": 10.190571314422693,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 25507
    },
    {
      "epoch": 10.190970834998003,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 25508
    },
    {
      "epoch": 10.191370355573312,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 25509
    },
    {
      "epoch": 10.191769876148621,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0488,
      "step": 25510
    },
    {
      "epoch": 10.192169396723932,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 25511
    },
    {
      "epoch": 10.19256891729924,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 25512
    },
    {
      "epoch": 10.192968437874551,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 25513
    },
    {
      "epoch": 10.19336795844986,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 25514
    },
    {
      "epoch": 10.193767479025169,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 25515
    },
    {
      "epoch": 10.19416699960048,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 25516
    },
    {
      "epoch": 10.194566520175789,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 25517
    },
    {
      "epoch": 10.1949660407511,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 25518
    },
    {
      "epoch": 10.195365561326408,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 25519
    },
    {
      "epoch": 10.195765081901717,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 25520
    },
    {
      "epoch": 10.196164602477028,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 25521
    },
    {
      "epoch": 10.196564123052337,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 25522
    },
    {
      "epoch": 10.196963643627647,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 25523
    },
    {
      "epoch": 10.197363164202956,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 25524
    },
    {
      "epoch": 10.197762684778265,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 25525
    },
    {
      "epoch": 10.198162205353576,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 25526
    },
    {
      "epoch": 10.198561725928885,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 25527
    },
    {
      "epoch": 10.198961246504195,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 25528
    },
    {
      "epoch": 10.199360767079504,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 25529
    },
    {
      "epoch": 10.199760287654815,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 25530
    },
    {
      "epoch": 10.200159808230124,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 25531
    },
    {
      "epoch": 10.200559328805433,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 25532
    },
    {
      "epoch": 10.200958849380743,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 25533
    },
    {
      "epoch": 10.201358369956052,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 25534
    },
    {
      "epoch": 10.201757890531363,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 25535
    },
    {
      "epoch": 10.202157411106672,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 25536
    },
    {
      "epoch": 10.20255693168198,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 25537
    },
    {
      "epoch": 10.202956452257292,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 25538
    },
    {
      "epoch": 10.2033559728326,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 25539
    },
    {
      "epoch": 10.203755493407911,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 25540
    },
    {
      "epoch": 10.20415501398322,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 25541
    },
    {
      "epoch": 10.204554534558529,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 25542
    },
    {
      "epoch": 10.20495405513384,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 25543
    },
    {
      "epoch": 10.205353575709148,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 25544
    },
    {
      "epoch": 10.20575309628446,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 25545
    },
    {
      "epoch": 10.206152616859768,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 25546
    },
    {
      "epoch": 10.206552137435079,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 25547
    },
    {
      "epoch": 10.206951658010388,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 25548
    },
    {
      "epoch": 10.207351178585697,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 25549
    },
    {
      "epoch": 10.207750699161007,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 25550
    },
    {
      "epoch": 10.208150219736316,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.049,
      "step": 25551
    },
    {
      "epoch": 10.208549740311627,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 25552
    },
    {
      "epoch": 10.208949260886936,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 25553
    },
    {
      "epoch": 10.209348781462245,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 25554
    },
    {
      "epoch": 10.209748302037555,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 25555
    },
    {
      "epoch": 10.210147822612864,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 25556
    },
    {
      "epoch": 10.210547343188175,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 25557
    },
    {
      "epoch": 10.210946863763484,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 25558
    },
    {
      "epoch": 10.211346384338793,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 25559
    },
    {
      "epoch": 10.211745904914103,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 25560
    },
    {
      "epoch": 10.212145425489412,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 25561
    },
    {
      "epoch": 10.212544946064723,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 25562
    },
    {
      "epoch": 10.212944466640032,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 25563
    },
    {
      "epoch": 10.213343987215342,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 25564
    },
    {
      "epoch": 10.213743507790651,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 25565
    },
    {
      "epoch": 10.21414302836596,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 25566
    },
    {
      "epoch": 10.214542548941271,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 25567
    },
    {
      "epoch": 10.21494206951658,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 25568
    },
    {
      "epoch": 10.21534159009189,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 25569
    },
    {
      "epoch": 10.2157411106672,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 25570
    },
    {
      "epoch": 10.216140631242508,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 25571
    },
    {
      "epoch": 10.216540151817819,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 25572
    },
    {
      "epoch": 10.216939672393128,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 25573
    },
    {
      "epoch": 10.217339192968439,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 25574
    },
    {
      "epoch": 10.217738713543747,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 25575
    },
    {
      "epoch": 10.218138234119056,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 25576
    },
    {
      "epoch": 10.218537754694367,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 25577
    },
    {
      "epoch": 10.218937275269676,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 25578
    },
    {
      "epoch": 10.219336795844987,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 25579
    },
    {
      "epoch": 10.219736316420295,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 25580
    },
    {
      "epoch": 10.220135836995606,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 25581
    },
    {
      "epoch": 10.220535357570915,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 25582
    },
    {
      "epoch": 10.220934878146224,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 25583
    },
    {
      "epoch": 10.221334398721535,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 25584
    },
    {
      "epoch": 10.221733919296843,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 25585
    },
    {
      "epoch": 10.222133439872154,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 25586
    },
    {
      "epoch": 10.222532960447463,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 25587
    },
    {
      "epoch": 10.222932481022772,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0579,
      "step": 25588
    },
    {
      "epoch": 10.223332001598083,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 25589
    },
    {
      "epoch": 10.223731522173392,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 25590
    },
    {
      "epoch": 10.224131042748702,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 25591
    },
    {
      "epoch": 10.224530563324011,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 25592
    },
    {
      "epoch": 10.22493008389932,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 25593
    },
    {
      "epoch": 10.22532960447463,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0511,
      "step": 25594
    },
    {
      "epoch": 10.22572912504994,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 25595
    },
    {
      "epoch": 10.22612864562525,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 25596
    },
    {
      "epoch": 10.22652816620056,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 25597
    },
    {
      "epoch": 10.22692768677587,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 25598
    },
    {
      "epoch": 10.227327207351179,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 25599
    },
    {
      "epoch": 10.227726727926488,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 25600
    },
    {
      "epoch": 10.228126248501798,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 25601
    },
    {
      "epoch": 10.228525769077107,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 25602
    },
    {
      "epoch": 10.228925289652418,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 25603
    },
    {
      "epoch": 10.229324810227727,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 25604
    },
    {
      "epoch": 10.229724330803036,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 25605
    },
    {
      "epoch": 10.230123851378346,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 25606
    },
    {
      "epoch": 10.230523371953655,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 25607
    },
    {
      "epoch": 10.230922892528966,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 25608
    },
    {
      "epoch": 10.231322413104275,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 25609
    },
    {
      "epoch": 10.231721933679584,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 25610
    },
    {
      "epoch": 10.232121454254894,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 25611
    },
    {
      "epoch": 10.232520974830203,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 25612
    },
    {
      "epoch": 10.232920495405514,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 25613
    },
    {
      "epoch": 10.233320015980823,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 25614
    },
    {
      "epoch": 10.233719536556134,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 25615
    },
    {
      "epoch": 10.234119057131442,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 25616
    },
    {
      "epoch": 10.234518577706751,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 25617
    },
    {
      "epoch": 10.234918098282062,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 25618
    },
    {
      "epoch": 10.235317618857371,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 25619
    },
    {
      "epoch": 10.235717139432682,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 25620
    },
    {
      "epoch": 10.23611666000799,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 25621
    },
    {
      "epoch": 10.2365161805833,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 25622
    },
    {
      "epoch": 10.23691570115861,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 25623
    },
    {
      "epoch": 10.237315221733919,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 25624
    },
    {
      "epoch": 10.23771474230923,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 25625
    },
    {
      "epoch": 10.238114262884539,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 25626
    },
    {
      "epoch": 10.238513783459847,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 25627
    },
    {
      "epoch": 10.238913304035158,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 25628
    },
    {
      "epoch": 10.239312824610467,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 25629
    },
    {
      "epoch": 10.239712345185778,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 25630
    },
    {
      "epoch": 10.240111865761087,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 25631
    },
    {
      "epoch": 10.240511386336395,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 25632
    },
    {
      "epoch": 10.240910906911706,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 25633
    },
    {
      "epoch": 10.241310427487015,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 25634
    },
    {
      "epoch": 10.241709948062326,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 25635
    },
    {
      "epoch": 10.242109468637635,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 25636
    },
    {
      "epoch": 10.242508989212945,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 25637
    },
    {
      "epoch": 10.242908509788254,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 25638
    },
    {
      "epoch": 10.243308030363563,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 25639
    },
    {
      "epoch": 10.243707550938874,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 25640
    },
    {
      "epoch": 10.244107071514183,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 25641
    },
    {
      "epoch": 10.244506592089493,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 25642
    },
    {
      "epoch": 10.244906112664802,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 25643
    },
    {
      "epoch": 10.245305633240111,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 25644
    },
    {
      "epoch": 10.245705153815422,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 25645
    },
    {
      "epoch": 10.24610467439073,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 25646
    },
    {
      "epoch": 10.246504194966041,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 25647
    },
    {
      "epoch": 10.24690371554135,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 25648
    },
    {
      "epoch": 10.24730323611666,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 25649
    },
    {
      "epoch": 10.24770275669197,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 25650
    },
    {
      "epoch": 10.248102277267279,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 25651
    },
    {
      "epoch": 10.24850179784259,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 25652
    },
    {
      "epoch": 10.248901318417898,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 25653
    },
    {
      "epoch": 10.249300838993209,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 25654
    },
    {
      "epoch": 10.249700359568518,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 25655
    },
    {
      "epoch": 10.250099880143827,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 25656
    },
    {
      "epoch": 10.250499400719137,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 25657
    },
    {
      "epoch": 10.250898921294446,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 25658
    },
    {
      "epoch": 10.251298441869757,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 25659
    },
    {
      "epoch": 10.251697962445066,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 25660
    },
    {
      "epoch": 10.252097483020375,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 25661
    },
    {
      "epoch": 10.252497003595685,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 25662
    },
    {
      "epoch": 10.252896524170994,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 25663
    },
    {
      "epoch": 10.253296044746305,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 25664
    },
    {
      "epoch": 10.253695565321614,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 25665
    },
    {
      "epoch": 10.254095085896923,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 25666
    },
    {
      "epoch": 10.254494606472234,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 25667
    },
    {
      "epoch": 10.254894127047542,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 25668
    },
    {
      "epoch": 10.255293647622853,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 25669
    },
    {
      "epoch": 10.255693168198162,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 25670
    },
    {
      "epoch": 10.256092688773473,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 25671
    },
    {
      "epoch": 10.256492209348782,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 25672
    },
    {
      "epoch": 10.25689172992409,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 25673
    },
    {
      "epoch": 10.257291250499401,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 25674
    },
    {
      "epoch": 10.25769077107471,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 25675
    },
    {
      "epoch": 10.25809029165002,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 25676
    },
    {
      "epoch": 10.25848981222533,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 25677
    },
    {
      "epoch": 10.258889332800639,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 25678
    },
    {
      "epoch": 10.25928885337595,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 25679
    },
    {
      "epoch": 10.259688373951258,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 25680
    },
    {
      "epoch": 10.260087894526569,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 25681
    },
    {
      "epoch": 10.260487415101878,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 25682
    },
    {
      "epoch": 10.260886935677187,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 25683
    },
    {
      "epoch": 10.261286456252497,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 25684
    },
    {
      "epoch": 10.261685976827806,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 25685
    },
    {
      "epoch": 10.262085497403117,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 25686
    },
    {
      "epoch": 10.262485017978426,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 25687
    },
    {
      "epoch": 10.262884538553735,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 25688
    },
    {
      "epoch": 10.263284059129045,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 25689
    },
    {
      "epoch": 10.263683579704354,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 25690
    },
    {
      "epoch": 10.264083100279665,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 25691
    },
    {
      "epoch": 10.264482620854974,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 25692
    },
    {
      "epoch": 10.264882141430284,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 25693
    },
    {
      "epoch": 10.265281662005593,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 25694
    },
    {
      "epoch": 10.265681182580902,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 25695
    },
    {
      "epoch": 10.266080703156213,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 25696
    },
    {
      "epoch": 10.266480223731522,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 25697
    },
    {
      "epoch": 10.266879744306832,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 25698
    },
    {
      "epoch": 10.267279264882141,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 25699
    },
    {
      "epoch": 10.26767878545745,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 25700
    },
    {
      "epoch": 10.268078306032761,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 25701
    },
    {
      "epoch": 10.26847782660807,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 25702
    },
    {
      "epoch": 10.26887734718338,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 25703
    },
    {
      "epoch": 10.26927686775869,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 25704
    },
    {
      "epoch": 10.269676388333998,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 25705
    },
    {
      "epoch": 10.270075908909309,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 25706
    },
    {
      "epoch": 10.270475429484618,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 25707
    },
    {
      "epoch": 10.270874950059929,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 25708
    },
    {
      "epoch": 10.271274470635237,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 25709
    },
    {
      "epoch": 10.271673991210548,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 25710
    },
    {
      "epoch": 10.272073511785857,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 25711
    },
    {
      "epoch": 10.272473032361166,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 25712
    },
    {
      "epoch": 10.272872552936477,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 25713
    },
    {
      "epoch": 10.273272073511786,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 25714
    },
    {
      "epoch": 10.273671594087096,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 25715
    },
    {
      "epoch": 10.274071114662405,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 25716
    },
    {
      "epoch": 10.274470635237714,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 25717
    },
    {
      "epoch": 10.274870155813025,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 25718
    },
    {
      "epoch": 10.275269676388334,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 25719
    },
    {
      "epoch": 10.275669196963644,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 25720
    },
    {
      "epoch": 10.276068717538953,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 25721
    },
    {
      "epoch": 10.276468238114262,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 25722
    },
    {
      "epoch": 10.276867758689573,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 25723
    },
    {
      "epoch": 10.277267279264882,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 25724
    },
    {
      "epoch": 10.277666799840192,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 25725
    },
    {
      "epoch": 10.278066320415501,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 25726
    },
    {
      "epoch": 10.278465840990812,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 25727
    },
    {
      "epoch": 10.27886536156612,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 25728
    },
    {
      "epoch": 10.27926488214143,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 25729
    },
    {
      "epoch": 10.27966440271674,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 25730
    },
    {
      "epoch": 10.28006392329205,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 25731
    },
    {
      "epoch": 10.28046344386736,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 25732
    },
    {
      "epoch": 10.280862964442669,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 25733
    },
    {
      "epoch": 10.281262485017978,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 25734
    },
    {
      "epoch": 10.281662005593288,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 25735
    },
    {
      "epoch": 10.282061526168597,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 25736
    },
    {
      "epoch": 10.282461046743908,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 25737
    },
    {
      "epoch": 10.282860567319217,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 25738
    },
    {
      "epoch": 10.283260087894526,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 25739
    },
    {
      "epoch": 10.283659608469836,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 25740
    },
    {
      "epoch": 10.284059129045145,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 25741
    },
    {
      "epoch": 10.284458649620456,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 25742
    },
    {
      "epoch": 10.284858170195765,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 25743
    },
    {
      "epoch": 10.285257690771076,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 25744
    },
    {
      "epoch": 10.285657211346384,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 25745
    },
    {
      "epoch": 10.286056731921693,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 25746
    },
    {
      "epoch": 10.286456252497004,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 25747
    },
    {
      "epoch": 10.286855773072313,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 25748
    },
    {
      "epoch": 10.287255293647624,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 25749
    },
    {
      "epoch": 10.287654814222932,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 25750
    },
    {
      "epoch": 10.288054334798241,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 25751
    },
    {
      "epoch": 10.288453855373552,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 25752
    },
    {
      "epoch": 10.288853375948861,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 25753
    },
    {
      "epoch": 10.289252896524172,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 25754
    },
    {
      "epoch": 10.28965241709948,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 25755
    },
    {
      "epoch": 10.29005193767479,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 25756
    },
    {
      "epoch": 10.2904514582501,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 25757
    },
    {
      "epoch": 10.290850978825409,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 25758
    },
    {
      "epoch": 10.29125049940072,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 25759
    },
    {
      "epoch": 10.291650019976029,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 25760
    },
    {
      "epoch": 10.29204954055134,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 25761
    },
    {
      "epoch": 10.292449061126648,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 25762
    },
    {
      "epoch": 10.292848581701957,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 25763
    },
    {
      "epoch": 10.293248102277268,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 25764
    },
    {
      "epoch": 10.293647622852577,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 25765
    },
    {
      "epoch": 10.294047143427887,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 25766
    },
    {
      "epoch": 10.294446664003196,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 25767
    },
    {
      "epoch": 10.294846184578505,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 25768
    },
    {
      "epoch": 10.295245705153816,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 25769
    },
    {
      "epoch": 10.295645225729125,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 25770
    },
    {
      "epoch": 10.296044746304435,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 25771
    },
    {
      "epoch": 10.296444266879744,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 25772
    },
    {
      "epoch": 10.296843787455053,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 25773
    },
    {
      "epoch": 10.297243308030364,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 25774
    },
    {
      "epoch": 10.297642828605673,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 25775
    },
    {
      "epoch": 10.298042349180983,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 25776
    },
    {
      "epoch": 10.298441869756292,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 25777
    },
    {
      "epoch": 10.298841390331603,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 25778
    },
    {
      "epoch": 10.299240910906912,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 25779
    },
    {
      "epoch": 10.29964043148222,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 25780
    },
    {
      "epoch": 10.300039952057531,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 25781
    },
    {
      "epoch": 10.30043947263284,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 25782
    },
    {
      "epoch": 10.300838993208151,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 25783
    },
    {
      "epoch": 10.30123851378346,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 25784
    },
    {
      "epoch": 10.301638034358769,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 25785
    },
    {
      "epoch": 10.30203755493408,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 25786
    },
    {
      "epoch": 10.302437075509388,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 25787
    },
    {
      "epoch": 10.302836596084699,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 25788
    },
    {
      "epoch": 10.303236116660008,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 25789
    },
    {
      "epoch": 10.303635637235317,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 25790
    },
    {
      "epoch": 10.304035157810628,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 25791
    },
    {
      "epoch": 10.304434678385936,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 25792
    },
    {
      "epoch": 10.304834198961247,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 25793
    },
    {
      "epoch": 10.305233719536556,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 25794
    },
    {
      "epoch": 10.305633240111867,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 25795
    },
    {
      "epoch": 10.306032760687176,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 25796
    },
    {
      "epoch": 10.306432281262484,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 25797
    },
    {
      "epoch": 10.306831801837795,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 25798
    },
    {
      "epoch": 10.307231322413104,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 25799
    },
    {
      "epoch": 10.307630842988415,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 25800
    },
    {
      "epoch": 10.308030363563724,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 25801
    },
    {
      "epoch": 10.308429884139032,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 25802
    },
    {
      "epoch": 10.308829404714343,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 25803
    },
    {
      "epoch": 10.309228925289652,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 25804
    },
    {
      "epoch": 10.309628445864963,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 25805
    },
    {
      "epoch": 10.310027966440272,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 25806
    },
    {
      "epoch": 10.31042748701558,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 25807
    },
    {
      "epoch": 10.310827007590891,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 25808
    },
    {
      "epoch": 10.3112265281662,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 25809
    },
    {
      "epoch": 10.31162604874151,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 25810
    },
    {
      "epoch": 10.31202556931682,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 25811
    },
    {
      "epoch": 10.31242508989213,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 25812
    },
    {
      "epoch": 10.31282461046744,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 25813
    },
    {
      "epoch": 10.313224131042748,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 25814
    },
    {
      "epoch": 10.313623651618059,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 25815
    },
    {
      "epoch": 10.314023172193368,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 25816
    },
    {
      "epoch": 10.314422692768678,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 25817
    },
    {
      "epoch": 10.314822213343987,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 25818
    },
    {
      "epoch": 10.315221733919296,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 25819
    },
    {
      "epoch": 10.315621254494607,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 25820
    },
    {
      "epoch": 10.316020775069916,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 25821
    },
    {
      "epoch": 10.316420295645226,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 25822
    },
    {
      "epoch": 10.316819816220535,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 25823
    },
    {
      "epoch": 10.317219336795844,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 25824
    },
    {
      "epoch": 10.317618857371155,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 25825
    },
    {
      "epoch": 10.318018377946464,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 25826
    },
    {
      "epoch": 10.318417898521774,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 25827
    },
    {
      "epoch": 10.318817419097083,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 25828
    },
    {
      "epoch": 10.319216939672392,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 25829
    },
    {
      "epoch": 10.319616460247703,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 25830
    },
    {
      "epoch": 10.320015980823012,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 25831
    },
    {
      "epoch": 10.320415501398323,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 25832
    },
    {
      "epoch": 10.320815021973631,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 25833
    },
    {
      "epoch": 10.321214542548942,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 25834
    },
    {
      "epoch": 10.321614063124251,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 25835
    },
    {
      "epoch": 10.32201358369956,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 25836
    },
    {
      "epoch": 10.32241310427487,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 25837
    },
    {
      "epoch": 10.32281262485018,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 25838
    },
    {
      "epoch": 10.32321214542549,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 25839
    },
    {
      "epoch": 10.323611666000799,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 25840
    },
    {
      "epoch": 10.324011186576108,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 25841
    },
    {
      "epoch": 10.324410707151419,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 25842
    },
    {
      "epoch": 10.324810227726728,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 25843
    },
    {
      "epoch": 10.325209748302038,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 25844
    },
    {
      "epoch": 10.325609268877347,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 25845
    },
    {
      "epoch": 10.326008789452656,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 25846
    },
    {
      "epoch": 10.326408310027967,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 25847
    },
    {
      "epoch": 10.326807830603276,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 25848
    },
    {
      "epoch": 10.327207351178586,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 25849
    },
    {
      "epoch": 10.327606871753895,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 25850
    },
    {
      "epoch": 10.328006392329206,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 25851
    },
    {
      "epoch": 10.328405912904515,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 25852
    },
    {
      "epoch": 10.328805433479824,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 25853
    },
    {
      "epoch": 10.329204954055134,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 25854
    },
    {
      "epoch": 10.329604474630443,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 25855
    },
    {
      "epoch": 10.330003995205754,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 25856
    },
    {
      "epoch": 10.330403515781063,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 25857
    },
    {
      "epoch": 10.330803036356372,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 25858
    },
    {
      "epoch": 10.331202556931682,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 25859
    },
    {
      "epoch": 10.331602077506991,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 25860
    },
    {
      "epoch": 10.332001598082302,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 25861
    },
    {
      "epoch": 10.33240111865761,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 25862
    },
    {
      "epoch": 10.33280063923292,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 25863
    },
    {
      "epoch": 10.33320015980823,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 25864
    },
    {
      "epoch": 10.33359968038354,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 25865
    },
    {
      "epoch": 10.33399920095885,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 25866
    },
    {
      "epoch": 10.334398721534159,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 25867
    },
    {
      "epoch": 10.334798242109468,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 25868
    },
    {
      "epoch": 10.335197762684778,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 25869
    },
    {
      "epoch": 10.335597283260087,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 25870
    },
    {
      "epoch": 10.335996803835398,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 25871
    },
    {
      "epoch": 10.336396324410707,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 25872
    },
    {
      "epoch": 10.336795844986018,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 25873
    },
    {
      "epoch": 10.337195365561326,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 25874
    },
    {
      "epoch": 10.337594886136635,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 25875
    },
    {
      "epoch": 10.337994406711946,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 25876
    },
    {
      "epoch": 10.338393927287255,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 25877
    },
    {
      "epoch": 10.338793447862566,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 25878
    },
    {
      "epoch": 10.339192968437874,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 25879
    },
    {
      "epoch": 10.339592489013183,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 25880
    },
    {
      "epoch": 10.339992009588494,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 25881
    },
    {
      "epoch": 10.340391530163803,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 25882
    },
    {
      "epoch": 10.340791050739114,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 25883
    },
    {
      "epoch": 10.341190571314423,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 25884
    },
    {
      "epoch": 10.341590091889731,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 25885
    },
    {
      "epoch": 10.341989612465042,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 25886
    },
    {
      "epoch": 10.342389133040351,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 25887
    },
    {
      "epoch": 10.342788653615662,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 25888
    },
    {
      "epoch": 10.34318817419097,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 25889
    },
    {
      "epoch": 10.343587694766281,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 25890
    },
    {
      "epoch": 10.34398721534159,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 25891
    },
    {
      "epoch": 10.344386735916899,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0535,
      "step": 25892
    },
    {
      "epoch": 10.34478625649221,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 25893
    },
    {
      "epoch": 10.345185777067519,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 25894
    },
    {
      "epoch": 10.34558529764283,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 25895
    },
    {
      "epoch": 10.345984818218138,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 25896
    },
    {
      "epoch": 10.346384338793447,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 25897
    },
    {
      "epoch": 10.346783859368758,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 25898
    },
    {
      "epoch": 10.347183379944067,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 25899
    },
    {
      "epoch": 10.347582900519377,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 25900
    },
    {
      "epoch": 10.347982421094686,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 25901
    },
    {
      "epoch": 10.348381941669995,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 25902
    },
    {
      "epoch": 10.348781462245306,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 25903
    },
    {
      "epoch": 10.349180982820615,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 25904
    },
    {
      "epoch": 10.349580503395925,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 25905
    },
    {
      "epoch": 10.349980023971234,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 25906
    },
    {
      "epoch": 10.350379544546545,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 25907
    },
    {
      "epoch": 10.350779065121854,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 25908
    },
    {
      "epoch": 10.351178585697163,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 25909
    },
    {
      "epoch": 10.351578106272473,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 25910
    },
    {
      "epoch": 10.351977626847782,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 25911
    },
    {
      "epoch": 10.352377147423093,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 25912
    },
    {
      "epoch": 10.352776667998402,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 25913
    },
    {
      "epoch": 10.35317618857371,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 25914
    },
    {
      "epoch": 10.353575709149021,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 25915
    },
    {
      "epoch": 10.35397522972433,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 25916
    },
    {
      "epoch": 10.354374750299641,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 25917
    },
    {
      "epoch": 10.35477427087495,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 25918
    },
    {
      "epoch": 10.355173791450259,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 25919
    },
    {
      "epoch": 10.35557331202557,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 25920
    },
    {
      "epoch": 10.355972832600878,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 25921
    },
    {
      "epoch": 10.356372353176189,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 25922
    },
    {
      "epoch": 10.356771873751498,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 25923
    },
    {
      "epoch": 10.357171394326809,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 25924
    },
    {
      "epoch": 10.357570914902118,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 25925
    },
    {
      "epoch": 10.357970435477426,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 25926
    },
    {
      "epoch": 10.358369956052737,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 25927
    },
    {
      "epoch": 10.358769476628046,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 25928
    },
    {
      "epoch": 10.359168997203357,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 25929
    },
    {
      "epoch": 10.359568517778666,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 25930
    },
    {
      "epoch": 10.359968038353975,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 25931
    },
    {
      "epoch": 10.360367558929285,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 25932
    },
    {
      "epoch": 10.360767079504594,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 25933
    },
    {
      "epoch": 10.361166600079905,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 25934
    },
    {
      "epoch": 10.361566120655214,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 25935
    },
    {
      "epoch": 10.361965641230523,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 25936
    },
    {
      "epoch": 10.362365161805833,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 25937
    },
    {
      "epoch": 10.362764682381142,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 25938
    },
    {
      "epoch": 10.363164202956453,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 25939
    },
    {
      "epoch": 10.363563723531762,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 25940
    },
    {
      "epoch": 10.363963244107072,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 25941
    },
    {
      "epoch": 10.364362764682381,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 25942
    },
    {
      "epoch": 10.36476228525769,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 25943
    },
    {
      "epoch": 10.365161805833,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 25944
    },
    {
      "epoch": 10.36556132640831,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 25945
    },
    {
      "epoch": 10.36596084698362,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 25946
    },
    {
      "epoch": 10.36636036755893,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 25947
    },
    {
      "epoch": 10.366759888134238,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 25948
    },
    {
      "epoch": 10.367159408709549,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 25949
    },
    {
      "epoch": 10.367558929284858,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 25950
    },
    {
      "epoch": 10.367958449860168,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 25951
    },
    {
      "epoch": 10.368357970435477,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 25952
    },
    {
      "epoch": 10.368757491010786,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 25953
    },
    {
      "epoch": 10.369157011586097,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 25954
    },
    {
      "epoch": 10.369556532161406,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 25955
    },
    {
      "epoch": 10.369956052736717,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 25956
    },
    {
      "epoch": 10.370355573312025,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 25957
    },
    {
      "epoch": 10.370755093887336,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 25958
    },
    {
      "epoch": 10.371154614462645,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 25959
    },
    {
      "epoch": 10.371554135037954,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 25960
    },
    {
      "epoch": 10.371953655613265,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 25961
    },
    {
      "epoch": 10.372353176188573,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 25962
    },
    {
      "epoch": 10.372752696763884,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 25963
    },
    {
      "epoch": 10.373152217339193,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 25964
    },
    {
      "epoch": 10.373551737914502,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 25965
    },
    {
      "epoch": 10.373951258489813,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0121,
      "step": 25966
    },
    {
      "epoch": 10.374350779065121,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 25967
    },
    {
      "epoch": 10.374750299640432,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 25968
    },
    {
      "epoch": 10.375149820215741,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 25969
    },
    {
      "epoch": 10.37554934079105,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 25970
    },
    {
      "epoch": 10.37594886136636,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 25971
    },
    {
      "epoch": 10.37634838194167,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 25972
    },
    {
      "epoch": 10.37674790251698,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 25973
    },
    {
      "epoch": 10.377147423092289,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 25974
    },
    {
      "epoch": 10.3775469436676,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 25975
    },
    {
      "epoch": 10.377946464242909,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 25976
    },
    {
      "epoch": 10.378345984818218,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 25977
    },
    {
      "epoch": 10.378745505393528,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 25978
    },
    {
      "epoch": 10.379145025968837,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 25979
    },
    {
      "epoch": 10.379544546544148,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 25980
    },
    {
      "epoch": 10.379944067119457,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 25981
    },
    {
      "epoch": 10.380343587694766,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 25982
    },
    {
      "epoch": 10.380743108270076,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 25983
    },
    {
      "epoch": 10.381142628845385,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 25984
    },
    {
      "epoch": 10.381542149420696,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 25985
    },
    {
      "epoch": 10.381941669996005,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 25986
    },
    {
      "epoch": 10.382341190571314,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 25987
    },
    {
      "epoch": 10.382740711146624,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 25988
    },
    {
      "epoch": 10.383140231721933,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 25989
    },
    {
      "epoch": 10.383539752297244,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 25990
    },
    {
      "epoch": 10.383939272872553,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 25991
    },
    {
      "epoch": 10.384338793447863,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 25992
    },
    {
      "epoch": 10.384738314023172,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 25993
    },
    {
      "epoch": 10.385137834598481,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 25994
    },
    {
      "epoch": 10.385537355173792,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 25995
    },
    {
      "epoch": 10.3859368757491,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 25996
    },
    {
      "epoch": 10.386336396324412,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 25997
    },
    {
      "epoch": 10.38673591689972,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 25998
    },
    {
      "epoch": 10.38713543747503,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 25999
    },
    {
      "epoch": 10.38753495805034,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 26000
    },
    {
      "epoch": 10.387934478625649,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 26001
    },
    {
      "epoch": 10.38833399920096,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 26002
    },
    {
      "epoch": 10.388733519776268,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 26003
    },
    {
      "epoch": 10.389133040351577,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 26004
    },
    {
      "epoch": 10.389532560926888,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 26005
    },
    {
      "epoch": 10.389932081502197,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 26006
    },
    {
      "epoch": 10.390331602077508,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 26007
    },
    {
      "epoch": 10.390731122652817,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 26008
    },
    {
      "epoch": 10.391130643228125,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 26009
    },
    {
      "epoch": 10.391530163803436,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 26010
    },
    {
      "epoch": 10.391929684378745,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 26011
    },
    {
      "epoch": 10.392329204954056,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 26012
    },
    {
      "epoch": 10.392728725529365,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 26013
    },
    {
      "epoch": 10.393128246104675,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 26014
    },
    {
      "epoch": 10.393527766679984,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 26015
    },
    {
      "epoch": 10.393927287255293,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 26016
    },
    {
      "epoch": 10.394326807830604,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 26017
    },
    {
      "epoch": 10.394726328405913,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 26018
    },
    {
      "epoch": 10.395125848981223,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 26019
    },
    {
      "epoch": 10.395525369556532,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 26020
    },
    {
      "epoch": 10.395924890131841,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 26021
    },
    {
      "epoch": 10.396324410707152,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 26022
    },
    {
      "epoch": 10.39672393128246,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 26023
    },
    {
      "epoch": 10.397123451857771,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 26024
    },
    {
      "epoch": 10.39752297243308,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 26025
    },
    {
      "epoch": 10.397922493008389,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 26026
    },
    {
      "epoch": 10.3983220135837,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 26027
    },
    {
      "epoch": 10.398721534159009,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 26028
    },
    {
      "epoch": 10.39912105473432,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 26029
    },
    {
      "epoch": 10.399520575309628,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 26030
    },
    {
      "epoch": 10.399920095884939,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 26031
    },
    {
      "epoch": 10.400319616460248,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 26032
    },
    {
      "epoch": 10.400719137035557,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 26033
    },
    {
      "epoch": 10.401118657610867,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 26034
    },
    {
      "epoch": 10.401518178186176,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 26035
    },
    {
      "epoch": 10.401917698761487,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 26036
    },
    {
      "epoch": 10.402317219336796,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 26037
    },
    {
      "epoch": 10.402716739912105,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 26038
    },
    {
      "epoch": 10.403116260487415,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 26039
    },
    {
      "epoch": 10.403515781062724,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 26040
    },
    {
      "epoch": 10.403915301638035,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 26041
    },
    {
      "epoch": 10.404314822213344,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 26042
    },
    {
      "epoch": 10.404714342788653,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 26043
    },
    {
      "epoch": 10.405113863363963,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 26044
    },
    {
      "epoch": 10.405513383939272,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 26045
    },
    {
      "epoch": 10.405912904514583,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 26046
    },
    {
      "epoch": 10.406312425089892,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 26047
    },
    {
      "epoch": 10.4067119456652,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 26048
    },
    {
      "epoch": 10.407111466240512,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 26049
    },
    {
      "epoch": 10.40751098681582,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 26050
    },
    {
      "epoch": 10.407910507391131,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 26051
    },
    {
      "epoch": 10.40831002796644,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 26052
    },
    {
      "epoch": 10.40870954854175,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 26053
    },
    {
      "epoch": 10.40910906911706,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 26054
    },
    {
      "epoch": 10.409508589692368,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 26055
    },
    {
      "epoch": 10.40990811026768,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 26056
    },
    {
      "epoch": 10.410307630842988,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 26057
    },
    {
      "epoch": 10.410707151418299,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 26058
    },
    {
      "epoch": 10.411106671993608,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 26059
    },
    {
      "epoch": 10.411506192568917,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 26060
    },
    {
      "epoch": 10.411905713144227,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 26061
    },
    {
      "epoch": 10.412305233719536,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 26062
    },
    {
      "epoch": 10.412704754294847,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 26063
    },
    {
      "epoch": 10.413104274870156,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 26064
    },
    {
      "epoch": 10.413503795445465,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 26065
    },
    {
      "epoch": 10.413903316020775,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 26066
    },
    {
      "epoch": 10.414302836596084,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 26067
    },
    {
      "epoch": 10.414702357171395,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 26068
    },
    {
      "epoch": 10.415101877746704,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 26069
    },
    {
      "epoch": 10.415501398322014,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 26070
    },
    {
      "epoch": 10.415900918897323,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 26071
    },
    {
      "epoch": 10.416300439472632,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 26072
    },
    {
      "epoch": 10.416699960047943,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 26073
    },
    {
      "epoch": 10.417099480623252,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 26074
    },
    {
      "epoch": 10.417499001198562,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 26075
    },
    {
      "epoch": 10.417898521773871,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 26076
    },
    {
      "epoch": 10.41829804234918,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 26077
    },
    {
      "epoch": 10.418697562924491,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 26078
    },
    {
      "epoch": 10.4190970834998,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 26079
    },
    {
      "epoch": 10.41949660407511,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 26080
    },
    {
      "epoch": 10.41989612465042,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 26081
    },
    {
      "epoch": 10.420295645225728,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 26082
    },
    {
      "epoch": 10.420695165801039,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 26083
    },
    {
      "epoch": 10.421094686376348,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 26084
    },
    {
      "epoch": 10.421494206951659,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 26085
    },
    {
      "epoch": 10.421893727526967,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 26086
    },
    {
      "epoch": 10.422293248102278,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 26087
    },
    {
      "epoch": 10.422692768677587,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 26088
    },
    {
      "epoch": 10.423092289252896,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 26089
    },
    {
      "epoch": 10.423491809828207,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 26090
    },
    {
      "epoch": 10.423891330403515,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 26091
    },
    {
      "epoch": 10.424290850978826,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 26092
    },
    {
      "epoch": 10.424690371554135,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 26093
    },
    {
      "epoch": 10.425089892129444,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 26094
    },
    {
      "epoch": 10.425489412704755,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 26095
    },
    {
      "epoch": 10.425888933280063,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 26096
    },
    {
      "epoch": 10.426288453855374,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 26097
    },
    {
      "epoch": 10.426687974430683,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 26098
    },
    {
      "epoch": 10.427087495005992,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 26099
    },
    {
      "epoch": 10.427487015581303,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 26100
    },
    {
      "epoch": 10.427886536156612,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 26101
    },
    {
      "epoch": 10.428286056731922,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 26102
    },
    {
      "epoch": 10.428685577307231,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 26103
    },
    {
      "epoch": 10.429085097882542,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 26104
    },
    {
      "epoch": 10.42948461845785,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 26105
    },
    {
      "epoch": 10.42988413903316,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 26106
    },
    {
      "epoch": 10.43028365960847,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 26107
    },
    {
      "epoch": 10.43068318018378,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 26108
    },
    {
      "epoch": 10.43108270075909,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 26109
    },
    {
      "epoch": 10.431482221334399,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 26110
    },
    {
      "epoch": 10.431881741909708,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 26111
    },
    {
      "epoch": 10.432281262485018,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 26112
    },
    {
      "epoch": 10.432680783060327,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 26113
    },
    {
      "epoch": 10.433080303635638,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 26114
    },
    {
      "epoch": 10.433479824210947,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 26115
    },
    {
      "epoch": 10.433879344786256,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 26116
    },
    {
      "epoch": 10.434278865361566,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 26117
    },
    {
      "epoch": 10.434678385936875,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 26118
    },
    {
      "epoch": 10.435077906512186,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 26119
    },
    {
      "epoch": 10.435477427087495,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 26120
    },
    {
      "epoch": 10.435876947662805,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 26121
    },
    {
      "epoch": 10.436276468238114,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 26122
    },
    {
      "epoch": 10.436675988813423,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 26123
    },
    {
      "epoch": 10.437075509388734,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 26124
    },
    {
      "epoch": 10.437475029964043,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 26125
    },
    {
      "epoch": 10.437874550539354,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 26126
    },
    {
      "epoch": 10.438274071114662,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 26127
    },
    {
      "epoch": 10.438673591689971,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 26128
    },
    {
      "epoch": 10.439073112265282,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 26129
    },
    {
      "epoch": 10.439472632840591,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 26130
    },
    {
      "epoch": 10.439872153415902,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 26131
    },
    {
      "epoch": 10.44027167399121,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 26132
    },
    {
      "epoch": 10.44067119456652,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 26133
    },
    {
      "epoch": 10.44107071514183,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 26134
    },
    {
      "epoch": 10.441470235717139,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 26135
    },
    {
      "epoch": 10.44186975629245,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 26136
    },
    {
      "epoch": 10.442269276867759,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 26137
    },
    {
      "epoch": 10.44266879744307,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 26138
    },
    {
      "epoch": 10.443068318018378,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 26139
    },
    {
      "epoch": 10.443467838593687,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 26140
    },
    {
      "epoch": 10.443867359168998,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 26141
    },
    {
      "epoch": 10.444266879744307,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 26142
    },
    {
      "epoch": 10.444666400319617,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 26143
    },
    {
      "epoch": 10.445065920894926,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 26144
    },
    {
      "epoch": 10.445465441470235,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 26145
    },
    {
      "epoch": 10.445864962045546,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 26146
    },
    {
      "epoch": 10.446264482620855,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 26147
    },
    {
      "epoch": 10.446664003196165,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 26148
    },
    {
      "epoch": 10.447063523771474,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 26149
    },
    {
      "epoch": 10.447463044346783,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 26150
    },
    {
      "epoch": 10.447862564922094,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 26151
    },
    {
      "epoch": 10.448262085497403,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 26152
    },
    {
      "epoch": 10.448661606072713,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 26153
    },
    {
      "epoch": 10.449061126648022,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 26154
    },
    {
      "epoch": 10.449460647223333,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 26155
    },
    {
      "epoch": 10.449860167798642,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 26156
    },
    {
      "epoch": 10.45025968837395,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 26157
    },
    {
      "epoch": 10.450659208949261,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 26158
    },
    {
      "epoch": 10.45105872952457,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 26159
    },
    {
      "epoch": 10.451458250099881,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 26160
    },
    {
      "epoch": 10.45185777067519,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 26161
    },
    {
      "epoch": 10.452257291250499,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 26162
    },
    {
      "epoch": 10.45265681182581,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 26163
    },
    {
      "epoch": 10.453056332401118,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 26164
    },
    {
      "epoch": 10.453455852976429,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 26165
    },
    {
      "epoch": 10.453855373551738,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 26166
    },
    {
      "epoch": 10.454254894127047,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 26167
    },
    {
      "epoch": 10.454654414702357,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 26168
    },
    {
      "epoch": 10.455053935277666,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 26169
    },
    {
      "epoch": 10.455453455852977,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 26170
    },
    {
      "epoch": 10.455852976428286,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 26171
    },
    {
      "epoch": 10.456252497003595,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 26172
    },
    {
      "epoch": 10.456652017578906,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 26173
    },
    {
      "epoch": 10.457051538154214,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 26174
    },
    {
      "epoch": 10.457451058729525,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 26175
    },
    {
      "epoch": 10.457850579304834,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 26176
    },
    {
      "epoch": 10.458250099880145,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 26177
    },
    {
      "epoch": 10.458649620455454,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 26178
    },
    {
      "epoch": 10.459049141030762,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 26179
    },
    {
      "epoch": 10.459448661606073,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 26180
    },
    {
      "epoch": 10.459848182181382,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 26181
    },
    {
      "epoch": 10.460247702756693,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 26182
    },
    {
      "epoch": 10.460647223332002,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 26183
    },
    {
      "epoch": 10.46104674390731,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 26184
    },
    {
      "epoch": 10.461446264482621,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 26185
    },
    {
      "epoch": 10.46184578505793,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 26186
    },
    {
      "epoch": 10.46224530563324,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 26187
    },
    {
      "epoch": 10.46264482620855,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 26188
    },
    {
      "epoch": 10.463044346783859,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 26189
    },
    {
      "epoch": 10.46344386735917,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 26190
    },
    {
      "epoch": 10.463843387934478,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 26191
    },
    {
      "epoch": 10.464242908509789,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 26192
    },
    {
      "epoch": 10.464642429085098,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 26193
    },
    {
      "epoch": 10.465041949660408,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 26194
    },
    {
      "epoch": 10.465441470235717,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 26195
    },
    {
      "epoch": 10.465840990811026,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 26196
    },
    {
      "epoch": 10.466240511386337,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 26197
    },
    {
      "epoch": 10.466640031961646,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 26198
    },
    {
      "epoch": 10.467039552536956,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 26199
    },
    {
      "epoch": 10.467439073112265,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 26200
    },
    {
      "epoch": 10.467838593687574,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 26201
    },
    {
      "epoch": 10.468238114262885,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 26202
    },
    {
      "epoch": 10.468637634838194,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 26203
    },
    {
      "epoch": 10.469037155413504,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 26204
    },
    {
      "epoch": 10.469436675988813,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 26205
    },
    {
      "epoch": 10.469836196564122,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 26206
    },
    {
      "epoch": 10.470235717139433,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 26207
    },
    {
      "epoch": 10.470635237714742,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 26208
    },
    {
      "epoch": 10.471034758290052,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 26209
    },
    {
      "epoch": 10.471434278865361,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 26210
    },
    {
      "epoch": 10.471833799440672,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 26211
    },
    {
      "epoch": 10.472233320015981,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 26212
    },
    {
      "epoch": 10.47263284059129,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 26213
    },
    {
      "epoch": 10.4730323611666,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 26214
    },
    {
      "epoch": 10.47343188174191,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 26215
    },
    {
      "epoch": 10.47383140231722,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 26216
    },
    {
      "epoch": 10.474230922892529,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 26217
    },
    {
      "epoch": 10.474630443467838,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 26218
    },
    {
      "epoch": 10.475029964043149,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 26219
    },
    {
      "epoch": 10.475429484618457,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 26220
    },
    {
      "epoch": 10.475829005193768,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 26221
    },
    {
      "epoch": 10.476228525769077,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 26222
    },
    {
      "epoch": 10.476628046344386,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 26223
    },
    {
      "epoch": 10.477027566919697,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 26224
    },
    {
      "epoch": 10.477427087495006,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 26225
    },
    {
      "epoch": 10.477826608070316,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 26226
    },
    {
      "epoch": 10.478226128645625,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 26227
    },
    {
      "epoch": 10.478625649220934,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 26228
    },
    {
      "epoch": 10.479025169796245,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 26229
    },
    {
      "epoch": 10.479424690371554,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 26230
    },
    {
      "epoch": 10.479824210946864,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 26231
    },
    {
      "epoch": 10.480223731522173,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 26232
    },
    {
      "epoch": 10.480623252097484,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 26233
    },
    {
      "epoch": 10.481022772672793,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 26234
    },
    {
      "epoch": 10.481422293248102,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 26235
    },
    {
      "epoch": 10.481821813823412,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 26236
    },
    {
      "epoch": 10.482221334398721,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 26237
    },
    {
      "epoch": 10.482620854974032,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 26238
    },
    {
      "epoch": 10.48302037554934,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 26239
    },
    {
      "epoch": 10.48341989612465,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 26240
    },
    {
      "epoch": 10.48381941669996,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 26241
    },
    {
      "epoch": 10.48421893727527,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 26242
    },
    {
      "epoch": 10.48461845785058,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 26243
    },
    {
      "epoch": 10.485017978425889,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 26244
    },
    {
      "epoch": 10.485417499001198,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 26245
    },
    {
      "epoch": 10.485817019576508,
      "grad_norm": 0.036865234375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 26246
    },
    {
      "epoch": 10.486216540151817,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 26247
    },
    {
      "epoch": 10.486616060727128,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 26248
    },
    {
      "epoch": 10.487015581302437,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 26249
    },
    {
      "epoch": 10.487415101877748,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 26250
    },
    {
      "epoch": 10.487814622453056,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 26251
    },
    {
      "epoch": 10.488214143028365,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 26252
    },
    {
      "epoch": 10.488613663603676,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 26253
    },
    {
      "epoch": 10.489013184178985,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 26254
    },
    {
      "epoch": 10.489412704754296,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 26255
    },
    {
      "epoch": 10.489812225329604,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 26256
    },
    {
      "epoch": 10.490211745904913,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 26257
    },
    {
      "epoch": 10.490611266480224,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 26258
    },
    {
      "epoch": 10.491010787055533,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 26259
    },
    {
      "epoch": 10.491410307630844,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 26260
    },
    {
      "epoch": 10.491809828206152,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 26261
    },
    {
      "epoch": 10.492209348781461,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 26262
    },
    {
      "epoch": 10.492608869356772,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 26263
    },
    {
      "epoch": 10.493008389932081,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 26264
    },
    {
      "epoch": 10.493407910507392,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 26265
    },
    {
      "epoch": 10.4938074310827,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 26266
    },
    {
      "epoch": 10.494206951658011,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 26267
    },
    {
      "epoch": 10.49460647223332,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 26268
    },
    {
      "epoch": 10.495005992808629,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 26269
    },
    {
      "epoch": 10.49540551338394,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 26270
    },
    {
      "epoch": 10.495805033959249,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 26271
    },
    {
      "epoch": 10.49620455453456,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 26272
    },
    {
      "epoch": 10.496604075109868,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 26273
    },
    {
      "epoch": 10.497003595685177,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 26274
    },
    {
      "epoch": 10.497403116260488,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 26275
    },
    {
      "epoch": 10.497802636835797,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 26276
    },
    {
      "epoch": 10.498202157411107,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 26277
    },
    {
      "epoch": 10.498601677986416,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 26278
    },
    {
      "epoch": 10.499001198561725,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 26279
    },
    {
      "epoch": 10.499400719137036,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 26280
    },
    {
      "epoch": 10.499800239712345,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 26281
    },
    {
      "epoch": 10.500199760287655,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 26282
    },
    {
      "epoch": 10.500599280862964,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 26283
    },
    {
      "epoch": 10.500998801438275,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 26284
    },
    {
      "epoch": 10.501398322013584,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 26285
    },
    {
      "epoch": 10.501797842588893,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 26286
    },
    {
      "epoch": 10.502197363164203,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 26287
    },
    {
      "epoch": 10.502596883739512,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 26288
    },
    {
      "epoch": 10.502996404314823,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 26289
    },
    {
      "epoch": 10.503395924890132,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 26290
    },
    {
      "epoch": 10.50379544546544,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 26291
    },
    {
      "epoch": 10.504194966040751,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 26292
    },
    {
      "epoch": 10.50459448661606,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 26293
    },
    {
      "epoch": 10.504994007191371,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 26294
    },
    {
      "epoch": 10.50539352776668,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 26295
    },
    {
      "epoch": 10.505793048341989,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 26296
    },
    {
      "epoch": 10.5061925689173,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 26297
    },
    {
      "epoch": 10.506592089492608,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 26298
    },
    {
      "epoch": 10.506991610067919,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 26299
    },
    {
      "epoch": 10.507391130643228,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 26300
    },
    {
      "epoch": 10.507790651218539,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 26301
    },
    {
      "epoch": 10.508190171793848,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 26302
    },
    {
      "epoch": 10.508589692369156,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 26303
    },
    {
      "epoch": 10.508989212944467,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 26304
    },
    {
      "epoch": 10.509388733519776,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 26305
    },
    {
      "epoch": 10.509788254095087,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 26306
    },
    {
      "epoch": 10.510187774670396,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 26307
    },
    {
      "epoch": 10.510587295245704,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 26308
    },
    {
      "epoch": 10.510986815821015,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 26309
    },
    {
      "epoch": 10.511386336396324,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 26310
    },
    {
      "epoch": 10.511785856971635,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 26311
    },
    {
      "epoch": 10.512185377546944,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 26312
    },
    {
      "epoch": 10.512584898122252,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 26313
    },
    {
      "epoch": 10.512984418697563,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 26314
    },
    {
      "epoch": 10.513383939272872,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 26315
    },
    {
      "epoch": 10.513783459848183,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 26316
    },
    {
      "epoch": 10.514182980423492,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 26317
    },
    {
      "epoch": 10.514582500998802,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 26318
    },
    {
      "epoch": 10.514982021574111,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 26319
    },
    {
      "epoch": 10.51538154214942,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 26320
    },
    {
      "epoch": 10.51578106272473,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 26321
    },
    {
      "epoch": 10.51618058330004,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 26322
    },
    {
      "epoch": 10.51658010387535,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 26323
    },
    {
      "epoch": 10.51697962445066,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 26324
    },
    {
      "epoch": 10.517379145025968,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 26325
    },
    {
      "epoch": 10.517778665601279,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 26326
    },
    {
      "epoch": 10.518178186176588,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 26327
    },
    {
      "epoch": 10.518577706751898,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 26328
    },
    {
      "epoch": 10.518977227327207,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 26329
    },
    {
      "epoch": 10.519376747902516,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 26330
    },
    {
      "epoch": 10.519776268477827,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 26331
    },
    {
      "epoch": 10.520175789053136,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 26332
    },
    {
      "epoch": 10.520575309628446,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 26333
    },
    {
      "epoch": 10.520974830203755,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 26334
    },
    {
      "epoch": 10.521374350779066,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0493,
      "step": 26335
    },
    {
      "epoch": 10.521773871354375,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 26336
    },
    {
      "epoch": 10.522173391929684,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 26337
    },
    {
      "epoch": 10.522572912504994,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 26338
    },
    {
      "epoch": 10.522972433080303,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 26339
    },
    {
      "epoch": 10.523371953655614,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 26340
    },
    {
      "epoch": 10.523771474230923,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 26341
    },
    {
      "epoch": 10.524170994806232,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 26342
    },
    {
      "epoch": 10.524570515381543,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 26343
    },
    {
      "epoch": 10.524970035956851,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 26344
    },
    {
      "epoch": 10.525369556532162,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 26345
    },
    {
      "epoch": 10.525769077107471,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 26346
    },
    {
      "epoch": 10.52616859768278,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 26347
    },
    {
      "epoch": 10.52656811825809,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 26348
    },
    {
      "epoch": 10.5269676388334,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 26349
    },
    {
      "epoch": 10.52736715940871,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 26350
    },
    {
      "epoch": 10.527766679984019,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0531,
      "step": 26351
    },
    {
      "epoch": 10.52816620055933,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 26352
    },
    {
      "epoch": 10.528565721134639,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 26353
    },
    {
      "epoch": 10.528965241709948,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 26354
    },
    {
      "epoch": 10.529364762285258,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 26355
    },
    {
      "epoch": 10.529764282860567,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 26356
    },
    {
      "epoch": 10.530163803435878,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 26357
    },
    {
      "epoch": 10.530563324011187,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 26358
    },
    {
      "epoch": 10.530962844586496,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 26359
    },
    {
      "epoch": 10.531362365161806,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 26360
    },
    {
      "epoch": 10.531761885737115,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 26361
    },
    {
      "epoch": 10.532161406312426,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 26362
    },
    {
      "epoch": 10.532560926887735,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 26363
    },
    {
      "epoch": 10.532960447463044,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 26364
    },
    {
      "epoch": 10.533359968038354,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 26365
    },
    {
      "epoch": 10.533759488613663,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 26366
    },
    {
      "epoch": 10.534159009188974,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 26367
    },
    {
      "epoch": 10.534558529764283,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 26368
    },
    {
      "epoch": 10.534958050339593,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 26369
    },
    {
      "epoch": 10.535357570914902,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 26370
    },
    {
      "epoch": 10.535757091490211,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 26371
    },
    {
      "epoch": 10.536156612065522,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 26372
    },
    {
      "epoch": 10.53655613264083,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 26373
    },
    {
      "epoch": 10.53695565321614,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 26374
    },
    {
      "epoch": 10.53735517379145,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 26375
    },
    {
      "epoch": 10.53775469436676,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 26376
    },
    {
      "epoch": 10.53815421494207,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 26377
    },
    {
      "epoch": 10.538553735517379,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 26378
    },
    {
      "epoch": 10.53895325609269,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 26379
    },
    {
      "epoch": 10.539352776667998,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 26380
    },
    {
      "epoch": 10.539752297243307,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 26381
    },
    {
      "epoch": 10.540151817818618,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 26382
    },
    {
      "epoch": 10.540551338393927,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 26383
    },
    {
      "epoch": 10.540950858969238,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0116,
      "step": 26384
    },
    {
      "epoch": 10.541350379544546,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 26385
    },
    {
      "epoch": 10.541749900119855,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 26386
    },
    {
      "epoch": 10.542149420695166,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 26387
    },
    {
      "epoch": 10.542548941270475,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 26388
    },
    {
      "epoch": 10.542948461845786,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 26389
    },
    {
      "epoch": 10.543347982421094,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 26390
    },
    {
      "epoch": 10.543747502996403,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 26391
    },
    {
      "epoch": 10.544147023571714,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 26392
    },
    {
      "epoch": 10.544546544147023,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 26393
    },
    {
      "epoch": 10.544946064722334,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 26394
    },
    {
      "epoch": 10.545345585297643,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 26395
    },
    {
      "epoch": 10.545745105872953,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 26396
    },
    {
      "epoch": 10.546144626448262,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 26397
    },
    {
      "epoch": 10.546544147023571,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 26398
    },
    {
      "epoch": 10.546943667598882,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 26399
    },
    {
      "epoch": 10.54734318817419,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 26400
    },
    {
      "epoch": 10.547742708749501,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 26401
    },
    {
      "epoch": 10.54814222932481,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 26402
    },
    {
      "epoch": 10.548541749900119,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 26403
    },
    {
      "epoch": 10.54894127047543,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 26404
    },
    {
      "epoch": 10.549340791050739,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 26405
    },
    {
      "epoch": 10.54974031162605,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 26406
    },
    {
      "epoch": 10.550139832201358,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 26407
    },
    {
      "epoch": 10.550539352776667,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 26408
    },
    {
      "epoch": 10.550938873351978,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 26409
    },
    {
      "epoch": 10.551338393927287,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 26410
    },
    {
      "epoch": 10.551737914502597,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 26411
    },
    {
      "epoch": 10.552137435077906,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 26412
    },
    {
      "epoch": 10.552536955653217,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 26413
    },
    {
      "epoch": 10.552936476228526,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 26414
    },
    {
      "epoch": 10.553335996803835,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 26415
    },
    {
      "epoch": 10.553735517379145,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 26416
    },
    {
      "epoch": 10.554135037954454,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 26417
    },
    {
      "epoch": 10.554534558529765,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 26418
    },
    {
      "epoch": 10.554934079105074,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 26419
    },
    {
      "epoch": 10.555333599680383,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 26420
    },
    {
      "epoch": 10.555733120255693,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 26421
    },
    {
      "epoch": 10.556132640831002,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 26422
    },
    {
      "epoch": 10.556532161406313,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 26423
    },
    {
      "epoch": 10.556931681981622,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 26424
    },
    {
      "epoch": 10.55733120255693,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 26425
    },
    {
      "epoch": 10.557730723132241,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 26426
    },
    {
      "epoch": 10.55813024370755,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 26427
    },
    {
      "epoch": 10.558529764282861,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 26428
    },
    {
      "epoch": 10.55892928485817,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 26429
    },
    {
      "epoch": 10.55932880543348,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 26430
    },
    {
      "epoch": 10.55972832600879,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 26431
    },
    {
      "epoch": 10.560127846584098,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 26432
    },
    {
      "epoch": 10.560527367159409,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 26433
    },
    {
      "epoch": 10.560926887734718,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 26434
    },
    {
      "epoch": 10.561326408310029,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 26435
    },
    {
      "epoch": 10.561725928885338,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 26436
    },
    {
      "epoch": 10.562125449460646,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 26437
    },
    {
      "epoch": 10.562524970035957,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 26438
    },
    {
      "epoch": 10.562924490611266,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 26439
    },
    {
      "epoch": 10.563324011186577,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 26440
    },
    {
      "epoch": 10.563723531761886,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 26441
    },
    {
      "epoch": 10.564123052337195,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 26442
    },
    {
      "epoch": 10.564522572912505,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 26443
    },
    {
      "epoch": 10.564922093487814,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 26444
    },
    {
      "epoch": 10.565321614063125,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 26445
    },
    {
      "epoch": 10.565721134638434,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 26446
    },
    {
      "epoch": 10.566120655213744,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 26447
    },
    {
      "epoch": 10.566520175789053,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 26448
    },
    {
      "epoch": 10.566919696364362,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 26449
    },
    {
      "epoch": 10.567319216939673,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 26450
    },
    {
      "epoch": 10.567718737514982,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 26451
    },
    {
      "epoch": 10.568118258090292,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 26452
    },
    {
      "epoch": 10.568517778665601,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 26453
    },
    {
      "epoch": 10.56891729924091,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 26454
    },
    {
      "epoch": 10.56931681981622,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 26455
    },
    {
      "epoch": 10.56971634039153,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 26456
    },
    {
      "epoch": 10.57011586096684,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0601,
      "step": 26457
    },
    {
      "epoch": 10.57051538154215,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 26458
    },
    {
      "epoch": 10.570914902117458,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 26459
    },
    {
      "epoch": 10.571314422692769,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 26460
    },
    {
      "epoch": 10.571713943268078,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 26461
    },
    {
      "epoch": 10.572113463843388,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 26462
    },
    {
      "epoch": 10.572512984418697,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 26463
    },
    {
      "epoch": 10.572912504994008,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 26464
    },
    {
      "epoch": 10.573312025569317,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 26465
    },
    {
      "epoch": 10.573711546144626,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 26466
    },
    {
      "epoch": 10.574111066719937,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 26467
    },
    {
      "epoch": 10.574510587295245,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 26468
    },
    {
      "epoch": 10.574910107870556,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 26469
    },
    {
      "epoch": 10.575309628445865,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 26470
    },
    {
      "epoch": 10.575709149021174,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 26471
    },
    {
      "epoch": 10.576108669596485,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 26472
    },
    {
      "epoch": 10.576508190171793,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 26473
    },
    {
      "epoch": 10.576907710747104,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 26474
    },
    {
      "epoch": 10.577307231322413,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 26475
    },
    {
      "epoch": 10.577706751897722,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 26476
    },
    {
      "epoch": 10.578106272473033,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 26477
    },
    {
      "epoch": 10.578505793048341,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 26478
    },
    {
      "epoch": 10.578905313623652,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 26479
    },
    {
      "epoch": 10.579304834198961,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0497,
      "step": 26480
    },
    {
      "epoch": 10.579704354774272,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 26481
    },
    {
      "epoch": 10.58010387534958,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 26482
    },
    {
      "epoch": 10.58050339592489,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 26483
    },
    {
      "epoch": 10.5809029165002,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 26484
    },
    {
      "epoch": 10.581302437075509,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 26485
    },
    {
      "epoch": 10.58170195765082,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 26486
    },
    {
      "epoch": 10.582101478226129,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 26487
    },
    {
      "epoch": 10.582500998801438,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 26488
    },
    {
      "epoch": 10.582900519376748,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 26489
    },
    {
      "epoch": 10.583300039952057,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 26490
    },
    {
      "epoch": 10.583699560527368,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 26491
    },
    {
      "epoch": 10.584099081102677,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 26492
    },
    {
      "epoch": 10.584498601677986,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 26493
    },
    {
      "epoch": 10.584898122253296,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 26494
    },
    {
      "epoch": 10.585297642828605,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 26495
    },
    {
      "epoch": 10.585697163403916,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 26496
    },
    {
      "epoch": 10.586096683979225,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 26497
    },
    {
      "epoch": 10.586496204554535,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 26498
    },
    {
      "epoch": 10.586895725129844,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 26499
    },
    {
      "epoch": 10.587295245705153,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 26500
    },
    {
      "epoch": 10.587694766280464,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 26501
    },
    {
      "epoch": 10.588094286855773,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 26502
    },
    {
      "epoch": 10.588493807431083,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 26503
    },
    {
      "epoch": 10.588893328006392,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 26504
    },
    {
      "epoch": 10.589292848581701,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 26505
    },
    {
      "epoch": 10.589692369157012,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 26506
    },
    {
      "epoch": 10.59009188973232,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 26507
    },
    {
      "epoch": 10.590491410307632,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 26508
    },
    {
      "epoch": 10.59089093088294,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 26509
    },
    {
      "epoch": 10.59129045145825,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 26510
    },
    {
      "epoch": 10.59168997203356,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 26511
    },
    {
      "epoch": 10.592089492608869,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 26512
    },
    {
      "epoch": 10.59248901318418,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 26513
    },
    {
      "epoch": 10.592888533759488,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 26514
    },
    {
      "epoch": 10.5932880543348,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 26515
    },
    {
      "epoch": 10.593687574910108,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 26516
    },
    {
      "epoch": 10.594087095485417,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 26517
    },
    {
      "epoch": 10.594486616060728,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 26518
    },
    {
      "epoch": 10.594886136636037,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 26519
    },
    {
      "epoch": 10.595285657211347,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 26520
    },
    {
      "epoch": 10.595685177786656,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 26521
    },
    {
      "epoch": 10.596084698361965,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 26522
    },
    {
      "epoch": 10.596484218937276,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 26523
    },
    {
      "epoch": 10.596883739512585,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 26524
    },
    {
      "epoch": 10.597283260087895,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 26525
    },
    {
      "epoch": 10.597682780663204,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 26526
    },
    {
      "epoch": 10.598082301238513,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 26527
    },
    {
      "epoch": 10.598481821813824,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 26528
    },
    {
      "epoch": 10.598881342389133,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 26529
    },
    {
      "epoch": 10.599280862964443,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 26530
    },
    {
      "epoch": 10.599680383539752,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 26531
    },
    {
      "epoch": 10.600079904115063,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 26532
    },
    {
      "epoch": 10.600479424690372,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 26533
    },
    {
      "epoch": 10.60087894526568,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 26534
    },
    {
      "epoch": 10.601278465840991,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 26535
    },
    {
      "epoch": 10.6016779864163,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 26536
    },
    {
      "epoch": 10.602077506991611,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 26537
    },
    {
      "epoch": 10.60247702756692,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 26538
    },
    {
      "epoch": 10.602876548142229,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 26539
    },
    {
      "epoch": 10.60327606871754,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 26540
    },
    {
      "epoch": 10.603675589292848,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 26541
    },
    {
      "epoch": 10.604075109868159,
      "grad_norm": 0.036865234375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 26542
    },
    {
      "epoch": 10.604474630443468,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 26543
    },
    {
      "epoch": 10.604874151018777,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 26544
    },
    {
      "epoch": 10.605273671594087,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 26545
    },
    {
      "epoch": 10.605673192169396,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 26546
    },
    {
      "epoch": 10.606072712744707,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 26547
    },
    {
      "epoch": 10.606472233320016,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 26548
    },
    {
      "epoch": 10.606871753895327,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 26549
    },
    {
      "epoch": 10.607271274470635,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 26550
    },
    {
      "epoch": 10.607670795045944,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 26551
    },
    {
      "epoch": 10.608070315621255,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 26552
    },
    {
      "epoch": 10.608469836196564,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 26553
    },
    {
      "epoch": 10.608869356771873,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 26554
    },
    {
      "epoch": 10.609268877347183,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 26555
    },
    {
      "epoch": 10.609668397922492,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 26556
    },
    {
      "epoch": 10.610067918497803,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 26557
    },
    {
      "epoch": 10.610467439073112,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 26558
    },
    {
      "epoch": 10.610866959648423,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 26559
    },
    {
      "epoch": 10.611266480223732,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 26560
    },
    {
      "epoch": 10.61166600079904,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 26561
    },
    {
      "epoch": 10.612065521374351,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 26562
    },
    {
      "epoch": 10.61246504194966,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 26563
    },
    {
      "epoch": 10.61286456252497,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 26564
    },
    {
      "epoch": 10.61326408310028,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 26565
    },
    {
      "epoch": 10.613663603675588,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 26566
    },
    {
      "epoch": 10.6140631242509,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 26567
    },
    {
      "epoch": 10.614462644826208,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 26568
    },
    {
      "epoch": 10.614862165401519,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 26569
    },
    {
      "epoch": 10.615261685976828,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 26570
    },
    {
      "epoch": 10.615661206552137,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 26571
    },
    {
      "epoch": 10.616060727127447,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 26572
    },
    {
      "epoch": 10.616460247702756,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 26573
    },
    {
      "epoch": 10.616859768278067,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 26574
    },
    {
      "epoch": 10.617259288853376,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 26575
    },
    {
      "epoch": 10.617658809428686,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 26576
    },
    {
      "epoch": 10.618058330003995,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 26577
    },
    {
      "epoch": 10.618457850579304,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 26578
    },
    {
      "epoch": 10.618857371154615,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 26579
    },
    {
      "epoch": 10.619256891729924,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 26580
    },
    {
      "epoch": 10.619656412305234,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 26581
    },
    {
      "epoch": 10.620055932880543,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 26582
    },
    {
      "epoch": 10.620455453455852,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 26583
    },
    {
      "epoch": 10.620854974031163,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 26584
    },
    {
      "epoch": 10.621254494606472,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 26585
    },
    {
      "epoch": 10.621654015181782,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 26586
    },
    {
      "epoch": 10.622053535757091,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 26587
    },
    {
      "epoch": 10.6224530563324,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 26588
    },
    {
      "epoch": 10.622852576907711,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 26589
    },
    {
      "epoch": 10.62325209748302,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 26590
    },
    {
      "epoch": 10.62365161805833,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0495,
      "step": 26591
    },
    {
      "epoch": 10.62405113863364,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 26592
    },
    {
      "epoch": 10.62445065920895,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 26593
    },
    {
      "epoch": 10.624850179784259,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 26594
    },
    {
      "epoch": 10.625249700359568,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 26595
    },
    {
      "epoch": 10.625649220934879,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 26596
    },
    {
      "epoch": 10.626048741510187,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 26597
    },
    {
      "epoch": 10.626448262085498,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 26598
    },
    {
      "epoch": 10.626847782660807,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 26599
    },
    {
      "epoch": 10.627247303236116,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 26600
    },
    {
      "epoch": 10.627646823811427,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 26601
    },
    {
      "epoch": 10.628046344386735,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 26602
    },
    {
      "epoch": 10.628445864962046,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 26603
    },
    {
      "epoch": 10.628845385537355,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 26604
    },
    {
      "epoch": 10.629244906112664,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 26605
    },
    {
      "epoch": 10.629644426687975,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 26606
    },
    {
      "epoch": 10.630043947263283,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 26607
    },
    {
      "epoch": 10.630443467838594,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 26608
    },
    {
      "epoch": 10.630842988413903,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 26609
    },
    {
      "epoch": 10.631242508989214,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 26610
    },
    {
      "epoch": 10.631642029564523,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 26611
    },
    {
      "epoch": 10.632041550139832,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 26612
    },
    {
      "epoch": 10.632441070715142,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 26613
    },
    {
      "epoch": 10.632840591290451,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 26614
    },
    {
      "epoch": 10.633240111865762,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 26615
    },
    {
      "epoch": 10.63363963244107,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 26616
    },
    {
      "epoch": 10.63403915301638,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 26617
    },
    {
      "epoch": 10.63443867359169,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 26618
    },
    {
      "epoch": 10.634838194167,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 26619
    },
    {
      "epoch": 10.63523771474231,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 26620
    },
    {
      "epoch": 10.635637235317619,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 26621
    },
    {
      "epoch": 10.636036755892928,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 26622
    },
    {
      "epoch": 10.636436276468238,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 26623
    },
    {
      "epoch": 10.636835797043547,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 26624
    },
    {
      "epoch": 10.637235317618858,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 26625
    },
    {
      "epoch": 10.637634838194167,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 26626
    },
    {
      "epoch": 10.638034358769477,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 26627
    },
    {
      "epoch": 10.638433879344786,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 26628
    },
    {
      "epoch": 10.638833399920095,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 26629
    },
    {
      "epoch": 10.639232920495406,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 26630
    },
    {
      "epoch": 10.639632441070715,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 26631
    },
    {
      "epoch": 10.640031961646025,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 26632
    },
    {
      "epoch": 10.640431482221334,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 26633
    },
    {
      "epoch": 10.640831002796643,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 26634
    },
    {
      "epoch": 10.641230523371954,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 26635
    },
    {
      "epoch": 10.641630043947263,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 26636
    },
    {
      "epoch": 10.642029564522574,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 26637
    },
    {
      "epoch": 10.642429085097882,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 26638
    },
    {
      "epoch": 10.642828605673191,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 26639
    },
    {
      "epoch": 10.643228126248502,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 26640
    },
    {
      "epoch": 10.643627646823811,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 26641
    },
    {
      "epoch": 10.644027167399122,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 26642
    },
    {
      "epoch": 10.64442668797443,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 26643
    },
    {
      "epoch": 10.644826208549741,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 26644
    },
    {
      "epoch": 10.64522572912505,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 26645
    },
    {
      "epoch": 10.645625249700359,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 26646
    },
    {
      "epoch": 10.64602477027567,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 26647
    },
    {
      "epoch": 10.646424290850979,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 26648
    },
    {
      "epoch": 10.64682381142629,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 26649
    },
    {
      "epoch": 10.647223332001598,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 26650
    },
    {
      "epoch": 10.647622852576907,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 26651
    },
    {
      "epoch": 10.648022373152218,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 26652
    },
    {
      "epoch": 10.648421893727527,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 26653
    },
    {
      "epoch": 10.648821414302837,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 26654
    },
    {
      "epoch": 10.649220934878146,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 26655
    },
    {
      "epoch": 10.649620455453455,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 26656
    },
    {
      "epoch": 10.650019976028766,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 26657
    },
    {
      "epoch": 10.650419496604075,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 26658
    },
    {
      "epoch": 10.650819017179385,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 26659
    },
    {
      "epoch": 10.651218537754694,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 26660
    },
    {
      "epoch": 10.651618058330005,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 26661
    },
    {
      "epoch": 10.652017578905314,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 26662
    },
    {
      "epoch": 10.652417099480623,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 26663
    },
    {
      "epoch": 10.652816620055933,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 26664
    },
    {
      "epoch": 10.653216140631242,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 26665
    },
    {
      "epoch": 10.653615661206553,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 26666
    },
    {
      "epoch": 10.654015181781862,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 26667
    },
    {
      "epoch": 10.65441470235717,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 26668
    },
    {
      "epoch": 10.654814222932481,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 26669
    },
    {
      "epoch": 10.65521374350779,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 26670
    },
    {
      "epoch": 10.655613264083101,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 26671
    },
    {
      "epoch": 10.65601278465841,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 26672
    },
    {
      "epoch": 10.656412305233719,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 26673
    },
    {
      "epoch": 10.65681182580903,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 26674
    },
    {
      "epoch": 10.657211346384338,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 26675
    },
    {
      "epoch": 10.657610866959649,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 26676
    },
    {
      "epoch": 10.658010387534958,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 26677
    },
    {
      "epoch": 10.658409908110269,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 26678
    },
    {
      "epoch": 10.658809428685577,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 26679
    },
    {
      "epoch": 10.659208949260886,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 26680
    },
    {
      "epoch": 10.659608469836197,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 26681
    },
    {
      "epoch": 10.660007990411506,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 26682
    },
    {
      "epoch": 10.660407510986817,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0465,
      "step": 26683
    },
    {
      "epoch": 10.660807031562126,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 26684
    },
    {
      "epoch": 10.661206552137434,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 26685
    },
    {
      "epoch": 10.661606072712745,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 26686
    },
    {
      "epoch": 10.662005593288054,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 26687
    },
    {
      "epoch": 10.662405113863365,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 26688
    },
    {
      "epoch": 10.662804634438674,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 26689
    },
    {
      "epoch": 10.663204155013982,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 26690
    },
    {
      "epoch": 10.663603675589293,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 26691
    },
    {
      "epoch": 10.664003196164602,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 26692
    },
    {
      "epoch": 10.664402716739913,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 26693
    },
    {
      "epoch": 10.664802237315222,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 26694
    },
    {
      "epoch": 10.665201757890532,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 26695
    },
    {
      "epoch": 10.665601278465841,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 26696
    },
    {
      "epoch": 10.66600079904115,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 26697
    },
    {
      "epoch": 10.66640031961646,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 26698
    },
    {
      "epoch": 10.66679984019177,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 26699
    },
    {
      "epoch": 10.66719936076708,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 26700
    },
    {
      "epoch": 10.66759888134239,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 26701
    },
    {
      "epoch": 10.667998401917698,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 26702
    },
    {
      "epoch": 10.668397922493009,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 26703
    },
    {
      "epoch": 10.668797443068318,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 26704
    },
    {
      "epoch": 10.669196963643628,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 26705
    },
    {
      "epoch": 10.669596484218937,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 26706
    },
    {
      "epoch": 10.669996004794246,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 26707
    },
    {
      "epoch": 10.670395525369557,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 26708
    },
    {
      "epoch": 10.670795045944866,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 26709
    },
    {
      "epoch": 10.671194566520176,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 26710
    },
    {
      "epoch": 10.671594087095485,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 26711
    },
    {
      "epoch": 10.671993607670796,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 26712
    },
    {
      "epoch": 10.672393128246105,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 26713
    },
    {
      "epoch": 10.672792648821414,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 26714
    },
    {
      "epoch": 10.673192169396724,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 26715
    },
    {
      "epoch": 10.673591689972033,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 26716
    },
    {
      "epoch": 10.673991210547344,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 26717
    },
    {
      "epoch": 10.674390731122653,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 26718
    },
    {
      "epoch": 10.674790251697962,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 26719
    },
    {
      "epoch": 10.675189772273272,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 26720
    },
    {
      "epoch": 10.675589292848581,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 26721
    },
    {
      "epoch": 10.675988813423892,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 26722
    },
    {
      "epoch": 10.676388333999201,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 26723
    },
    {
      "epoch": 10.67678785457451,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 26724
    },
    {
      "epoch": 10.67718737514982,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 26725
    },
    {
      "epoch": 10.67758689572513,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 26726
    },
    {
      "epoch": 10.67798641630044,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 26727
    },
    {
      "epoch": 10.678385936875749,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 26728
    },
    {
      "epoch": 10.67878545745106,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 26729
    },
    {
      "epoch": 10.679184978026369,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 26730
    },
    {
      "epoch": 10.679584498601677,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 26731
    },
    {
      "epoch": 10.679984019176988,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 26732
    },
    {
      "epoch": 10.680383539752297,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 26733
    },
    {
      "epoch": 10.680783060327606,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 26734
    },
    {
      "epoch": 10.681182580902917,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 26735
    },
    {
      "epoch": 10.681582101478226,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 26736
    },
    {
      "epoch": 10.681981622053536,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 26737
    },
    {
      "epoch": 10.682381142628845,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 26738
    },
    {
      "epoch": 10.682780663204156,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 26739
    },
    {
      "epoch": 10.683180183779465,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 26740
    },
    {
      "epoch": 10.683579704354774,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 26741
    },
    {
      "epoch": 10.683979224930084,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 26742
    },
    {
      "epoch": 10.684378745505393,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 26743
    },
    {
      "epoch": 10.684778266080704,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 26744
    },
    {
      "epoch": 10.685177786656013,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 26745
    },
    {
      "epoch": 10.685577307231322,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 26746
    },
    {
      "epoch": 10.685976827806632,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 26747
    },
    {
      "epoch": 10.686376348381941,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 26748
    },
    {
      "epoch": 10.686775868957252,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 26749
    },
    {
      "epoch": 10.68717538953256,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 26750
    },
    {
      "epoch": 10.68757491010787,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 26751
    },
    {
      "epoch": 10.68797443068318,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 26752
    },
    {
      "epoch": 10.68837395125849,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 26753
    },
    {
      "epoch": 10.6887734718338,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 26754
    },
    {
      "epoch": 10.689172992409109,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 26755
    },
    {
      "epoch": 10.68957251298442,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 26756
    },
    {
      "epoch": 10.689972033559728,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 26757
    },
    {
      "epoch": 10.690371554135037,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 26758
    },
    {
      "epoch": 10.690771074710348,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 26759
    },
    {
      "epoch": 10.691170595285657,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 26760
    },
    {
      "epoch": 10.691570115860968,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 26761
    },
    {
      "epoch": 10.691969636436276,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 26762
    },
    {
      "epoch": 10.692369157011585,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 26763
    },
    {
      "epoch": 10.692768677586896,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 26764
    },
    {
      "epoch": 10.693168198162205,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 26765
    },
    {
      "epoch": 10.693567718737516,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 26766
    },
    {
      "epoch": 10.693967239312824,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 26767
    },
    {
      "epoch": 10.694366759888133,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 26768
    },
    {
      "epoch": 10.694766280463444,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 26769
    },
    {
      "epoch": 10.695165801038753,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 26770
    },
    {
      "epoch": 10.695565321614064,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 26771
    },
    {
      "epoch": 10.695964842189372,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 26772
    },
    {
      "epoch": 10.696364362764683,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 26773
    },
    {
      "epoch": 10.696763883339992,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 26774
    },
    {
      "epoch": 10.697163403915301,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 26775
    },
    {
      "epoch": 10.697562924490612,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 26776
    },
    {
      "epoch": 10.69796244506592,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 26777
    },
    {
      "epoch": 10.698361965641231,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 26778
    },
    {
      "epoch": 10.69876148621654,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 26779
    },
    {
      "epoch": 10.699161006791849,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 26780
    },
    {
      "epoch": 10.69956052736716,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 26781
    },
    {
      "epoch": 10.699960047942469,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 26782
    },
    {
      "epoch": 10.70035956851778,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 26783
    },
    {
      "epoch": 10.700759089093088,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 26784
    },
    {
      "epoch": 10.701158609668397,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 26785
    },
    {
      "epoch": 10.701558130243708,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 26786
    },
    {
      "epoch": 10.701957650819017,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 26787
    },
    {
      "epoch": 10.702357171394327,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 26788
    },
    {
      "epoch": 10.702756691969636,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 26789
    },
    {
      "epoch": 10.703156212544947,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 26790
    },
    {
      "epoch": 10.703555733120256,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 26791
    },
    {
      "epoch": 10.703955253695565,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 26792
    },
    {
      "epoch": 10.704354774270875,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 26793
    },
    {
      "epoch": 10.704754294846184,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 26794
    },
    {
      "epoch": 10.705153815421495,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 26795
    },
    {
      "epoch": 10.705553335996804,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0127,
      "step": 26796
    },
    {
      "epoch": 10.705952856572113,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 26797
    },
    {
      "epoch": 10.706352377147423,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 26798
    },
    {
      "epoch": 10.706751897722732,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 26799
    },
    {
      "epoch": 10.707151418298043,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 26800
    },
    {
      "epoch": 10.707550938873352,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 26801
    },
    {
      "epoch": 10.70795045944866,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 26802
    },
    {
      "epoch": 10.708349980023971,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 26803
    },
    {
      "epoch": 10.70874950059928,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 26804
    },
    {
      "epoch": 10.709149021174591,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 26805
    },
    {
      "epoch": 10.7095485417499,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 26806
    },
    {
      "epoch": 10.70994806232521,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 26807
    },
    {
      "epoch": 10.71034758290052,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 26808
    },
    {
      "epoch": 10.710747103475828,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 26809
    },
    {
      "epoch": 10.711146624051139,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 26810
    },
    {
      "epoch": 10.711546144626448,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 26811
    },
    {
      "epoch": 10.711945665201759,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 26812
    },
    {
      "epoch": 10.712345185777068,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 26813
    },
    {
      "epoch": 10.712744706352376,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 26814
    },
    {
      "epoch": 10.713144226927687,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 26815
    },
    {
      "epoch": 10.713543747502996,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 26816
    },
    {
      "epoch": 10.713943268078307,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 26817
    },
    {
      "epoch": 10.714342788653616,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 26818
    },
    {
      "epoch": 10.714742309228924,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 26819
    },
    {
      "epoch": 10.715141829804235,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 26820
    },
    {
      "epoch": 10.715541350379544,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 26821
    },
    {
      "epoch": 10.715940870954855,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 26822
    },
    {
      "epoch": 10.716340391530164,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 26823
    },
    {
      "epoch": 10.716739912105474,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 26824
    },
    {
      "epoch": 10.717139432680783,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 26825
    },
    {
      "epoch": 10.717538953256092,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 26826
    },
    {
      "epoch": 10.717938473831403,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 26827
    },
    {
      "epoch": 10.718337994406712,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 26828
    },
    {
      "epoch": 10.718737514982022,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 26829
    },
    {
      "epoch": 10.719137035557331,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 26830
    },
    {
      "epoch": 10.71953655613264,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 26831
    },
    {
      "epoch": 10.71993607670795,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 26832
    },
    {
      "epoch": 10.72033559728326,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 26833
    },
    {
      "epoch": 10.72073511785857,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 26834
    },
    {
      "epoch": 10.72113463843388,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 26835
    },
    {
      "epoch": 10.721534159009188,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 26836
    },
    {
      "epoch": 10.721933679584499,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 26837
    },
    {
      "epoch": 10.722333200159808,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 26838
    },
    {
      "epoch": 10.722732720735118,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 26839
    },
    {
      "epoch": 10.723132241310427,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 26840
    },
    {
      "epoch": 10.723531761885738,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 26841
    },
    {
      "epoch": 10.723931282461047,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 26842
    },
    {
      "epoch": 10.724330803036356,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 26843
    },
    {
      "epoch": 10.724730323611666,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 26844
    },
    {
      "epoch": 10.725129844186975,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 26845
    },
    {
      "epoch": 10.725529364762286,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 26846
    },
    {
      "epoch": 10.725928885337595,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0092,
      "step": 26847
    },
    {
      "epoch": 10.726328405912904,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 26848
    },
    {
      "epoch": 10.726727926488214,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 26849
    },
    {
      "epoch": 10.727127447063523,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 26850
    },
    {
      "epoch": 10.727526967638834,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 26851
    },
    {
      "epoch": 10.727926488214143,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 26852
    },
    {
      "epoch": 10.728326008789452,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 26853
    },
    {
      "epoch": 10.728725529364763,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 26854
    },
    {
      "epoch": 10.729125049940071,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 26855
    },
    {
      "epoch": 10.729524570515382,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 26856
    },
    {
      "epoch": 10.729924091090691,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 26857
    },
    {
      "epoch": 10.730323611666002,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 26858
    },
    {
      "epoch": 10.73072313224131,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 26859
    },
    {
      "epoch": 10.73112265281662,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 26860
    },
    {
      "epoch": 10.73152217339193,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 26861
    },
    {
      "epoch": 10.731921693967239,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 26862
    },
    {
      "epoch": 10.73232121454255,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 26863
    },
    {
      "epoch": 10.732720735117859,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 26864
    },
    {
      "epoch": 10.733120255693168,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 26865
    },
    {
      "epoch": 10.733519776268478,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 26866
    },
    {
      "epoch": 10.733919296843787,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 26867
    },
    {
      "epoch": 10.734318817419098,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 26868
    },
    {
      "epoch": 10.734718337994407,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 26869
    },
    {
      "epoch": 10.735117858569716,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 26870
    },
    {
      "epoch": 10.735517379145026,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 26871
    },
    {
      "epoch": 10.735916899720335,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 26872
    },
    {
      "epoch": 10.736316420295646,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 26873
    },
    {
      "epoch": 10.736715940870955,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 26874
    },
    {
      "epoch": 10.737115461446265,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 26875
    },
    {
      "epoch": 10.737514982021574,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 26876
    },
    {
      "epoch": 10.737914502596883,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 26877
    },
    {
      "epoch": 10.738314023172194,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 26878
    },
    {
      "epoch": 10.738713543747503,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 26879
    },
    {
      "epoch": 10.739113064322813,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 26880
    },
    {
      "epoch": 10.739512584898122,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 26881
    },
    {
      "epoch": 10.739912105473431,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 26882
    },
    {
      "epoch": 10.740311626048742,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 26883
    },
    {
      "epoch": 10.74071114662405,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 26884
    },
    {
      "epoch": 10.741110667199361,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 26885
    },
    {
      "epoch": 10.74151018777467,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 26886
    },
    {
      "epoch": 10.74190970834998,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 26887
    },
    {
      "epoch": 10.74230922892529,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 26888
    },
    {
      "epoch": 10.742708749500599,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 26889
    },
    {
      "epoch": 10.74310827007591,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 26890
    },
    {
      "epoch": 10.743507790651218,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 26891
    },
    {
      "epoch": 10.743907311226529,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 26892
    },
    {
      "epoch": 10.744306831801838,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 26893
    },
    {
      "epoch": 10.744706352377147,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 26894
    },
    {
      "epoch": 10.745105872952458,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 26895
    },
    {
      "epoch": 10.745505393527766,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 26896
    },
    {
      "epoch": 10.745904914103077,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 26897
    },
    {
      "epoch": 10.746304434678386,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 26898
    },
    {
      "epoch": 10.746703955253695,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 26899
    },
    {
      "epoch": 10.747103475829006,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 26900
    },
    {
      "epoch": 10.747502996404315,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 26901
    },
    {
      "epoch": 10.747902516979625,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 26902
    },
    {
      "epoch": 10.748302037554934,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 26903
    },
    {
      "epoch": 10.748701558130243,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 26904
    },
    {
      "epoch": 10.749101078705554,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 26905
    },
    {
      "epoch": 10.749500599280863,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 26906
    },
    {
      "epoch": 10.749900119856173,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 26907
    },
    {
      "epoch": 10.750299640431482,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 26908
    },
    {
      "epoch": 10.750699161006793,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 26909
    },
    {
      "epoch": 10.751098681582102,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 26910
    },
    {
      "epoch": 10.75149820215741,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 26911
    },
    {
      "epoch": 10.751897722732721,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 26912
    },
    {
      "epoch": 10.75229724330803,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 26913
    },
    {
      "epoch": 10.752696763883339,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 26914
    },
    {
      "epoch": 10.75309628445865,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 26915
    },
    {
      "epoch": 10.753495805033959,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 26916
    },
    {
      "epoch": 10.75389532560927,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 26917
    },
    {
      "epoch": 10.754294846184578,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 26918
    },
    {
      "epoch": 10.754694366759889,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 26919
    },
    {
      "epoch": 10.755093887335198,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 26920
    },
    {
      "epoch": 10.755493407910507,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 26921
    },
    {
      "epoch": 10.755892928485817,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 26922
    },
    {
      "epoch": 10.756292449061126,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 26923
    },
    {
      "epoch": 10.756691969636437,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 26924
    },
    {
      "epoch": 10.757091490211746,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 26925
    },
    {
      "epoch": 10.757491010787055,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 26926
    },
    {
      "epoch": 10.757890531362365,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 26927
    },
    {
      "epoch": 10.758290051937674,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 26928
    },
    {
      "epoch": 10.758689572512985,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 26929
    },
    {
      "epoch": 10.759089093088294,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 26930
    },
    {
      "epoch": 10.759488613663603,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 26931
    },
    {
      "epoch": 10.759888134238913,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 26932
    },
    {
      "epoch": 10.760287654814222,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 26933
    },
    {
      "epoch": 10.760687175389533,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 26934
    },
    {
      "epoch": 10.761086695964842,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 26935
    },
    {
      "epoch": 10.761486216540153,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 26936
    },
    {
      "epoch": 10.761885737115461,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 26937
    },
    {
      "epoch": 10.76228525769077,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 26938
    },
    {
      "epoch": 10.762684778266081,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 26939
    },
    {
      "epoch": 10.76308429884139,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 26940
    },
    {
      "epoch": 10.7634838194167,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 26941
    },
    {
      "epoch": 10.76388333999201,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 26942
    },
    {
      "epoch": 10.764282860567318,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 26943
    },
    {
      "epoch": 10.764682381142629,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 26944
    },
    {
      "epoch": 10.765081901717938,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 26945
    },
    {
      "epoch": 10.765481422293249,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 26946
    },
    {
      "epoch": 10.765880942868558,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 26947
    },
    {
      "epoch": 10.766280463443866,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 26948
    },
    {
      "epoch": 10.766679984019177,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 26949
    },
    {
      "epoch": 10.767079504594486,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 26950
    },
    {
      "epoch": 10.767479025169797,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 26951
    },
    {
      "epoch": 10.767878545745106,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 26952
    },
    {
      "epoch": 10.768278066320416,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 26953
    },
    {
      "epoch": 10.768677586895725,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 26954
    },
    {
      "epoch": 10.769077107471034,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 26955
    },
    {
      "epoch": 10.769476628046345,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 26956
    },
    {
      "epoch": 10.769876148621654,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 26957
    },
    {
      "epoch": 10.770275669196964,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 26958
    },
    {
      "epoch": 10.770675189772273,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 26959
    },
    {
      "epoch": 10.771074710347582,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 26960
    },
    {
      "epoch": 10.771474230922893,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 26961
    },
    {
      "epoch": 10.771873751498202,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 26962
    },
    {
      "epoch": 10.772273272073512,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 26963
    },
    {
      "epoch": 10.772672792648821,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 26964
    },
    {
      "epoch": 10.77307231322413,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 26965
    },
    {
      "epoch": 10.77347183379944,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 26966
    },
    {
      "epoch": 10.77387135437475,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 26967
    },
    {
      "epoch": 10.77427087495006,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 26968
    },
    {
      "epoch": 10.77467039552537,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 26969
    },
    {
      "epoch": 10.77506991610068,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 26970
    },
    {
      "epoch": 10.775469436675989,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 26971
    },
    {
      "epoch": 10.775868957251298,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 26972
    },
    {
      "epoch": 10.776268477826608,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 26973
    },
    {
      "epoch": 10.776667998401917,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 26974
    },
    {
      "epoch": 10.777067518977228,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 26975
    },
    {
      "epoch": 10.777467039552537,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 26976
    },
    {
      "epoch": 10.777866560127846,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 26977
    },
    {
      "epoch": 10.778266080703157,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 26978
    },
    {
      "epoch": 10.778665601278465,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 26979
    },
    {
      "epoch": 10.779065121853776,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 26980
    },
    {
      "epoch": 10.779464642429085,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 26981
    },
    {
      "epoch": 10.779864163004394,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 26982
    },
    {
      "epoch": 10.780263683579705,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 26983
    },
    {
      "epoch": 10.780663204155013,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 26984
    },
    {
      "epoch": 10.781062724730324,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 26985
    },
    {
      "epoch": 10.781462245305633,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 26986
    },
    {
      "epoch": 10.781861765880944,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 26987
    },
    {
      "epoch": 10.782261286456253,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 26988
    },
    {
      "epoch": 10.782660807031561,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 26989
    },
    {
      "epoch": 10.783060327606872,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 26990
    },
    {
      "epoch": 10.783459848182181,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 26991
    },
    {
      "epoch": 10.783859368757492,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 26992
    },
    {
      "epoch": 10.7842588893328,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 26993
    },
    {
      "epoch": 10.78465840990811,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 26994
    },
    {
      "epoch": 10.78505793048342,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 26995
    },
    {
      "epoch": 10.785457451058729,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 26996
    },
    {
      "epoch": 10.78585697163404,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 26997
    },
    {
      "epoch": 10.786256492209349,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 26998
    },
    {
      "epoch": 10.786656012784658,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 26999
    },
    {
      "epoch": 10.787055533359968,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 27000
    },
    {
      "epoch": 10.787455053935277,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 27001
    },
    {
      "epoch": 10.787854574510588,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 27002
    },
    {
      "epoch": 10.788254095085897,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 27003
    },
    {
      "epoch": 10.788653615661207,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 27004
    },
    {
      "epoch": 10.789053136236516,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 27005
    },
    {
      "epoch": 10.789452656811825,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 27006
    },
    {
      "epoch": 10.789852177387136,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0517,
      "step": 27007
    },
    {
      "epoch": 10.790251697962445,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 27008
    },
    {
      "epoch": 10.790651218537755,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 27009
    },
    {
      "epoch": 10.791050739113064,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 27010
    },
    {
      "epoch": 10.791450259688373,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 27011
    },
    {
      "epoch": 10.791849780263684,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 27012
    },
    {
      "epoch": 10.792249300838993,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 27013
    },
    {
      "epoch": 10.792648821414303,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 27014
    },
    {
      "epoch": 10.793048341989612,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 27015
    },
    {
      "epoch": 10.793447862564921,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 27016
    },
    {
      "epoch": 10.793847383140232,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 27017
    },
    {
      "epoch": 10.79424690371554,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 27018
    },
    {
      "epoch": 10.794646424290852,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 27019
    },
    {
      "epoch": 10.79504594486616,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 27020
    },
    {
      "epoch": 10.795445465441471,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 27021
    },
    {
      "epoch": 10.79584498601678,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 27022
    },
    {
      "epoch": 10.796244506592089,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 27023
    },
    {
      "epoch": 10.7966440271674,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 27024
    },
    {
      "epoch": 10.797043547742708,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 27025
    },
    {
      "epoch": 10.79744306831802,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 27026
    },
    {
      "epoch": 10.797842588893328,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0533,
      "step": 27027
    },
    {
      "epoch": 10.798242109468637,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 27028
    },
    {
      "epoch": 10.798641630043948,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 27029
    },
    {
      "epoch": 10.799041150619257,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 27030
    },
    {
      "epoch": 10.799440671194567,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 27031
    },
    {
      "epoch": 10.799840191769876,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 27032
    },
    {
      "epoch": 10.800239712345185,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 27033
    },
    {
      "epoch": 10.800639232920496,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 27034
    },
    {
      "epoch": 10.801038753495805,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 27035
    },
    {
      "epoch": 10.801438274071115,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 27036
    },
    {
      "epoch": 10.801837794646424,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 27037
    },
    {
      "epoch": 10.802237315221735,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 27038
    },
    {
      "epoch": 10.802636835797044,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 27039
    },
    {
      "epoch": 10.803036356372353,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 27040
    },
    {
      "epoch": 10.803435876947663,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 27041
    },
    {
      "epoch": 10.803835397522972,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 27042
    },
    {
      "epoch": 10.804234918098283,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 27043
    },
    {
      "epoch": 10.804634438673592,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 27044
    },
    {
      "epoch": 10.8050339592489,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 27045
    },
    {
      "epoch": 10.805433479824211,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 27046
    },
    {
      "epoch": 10.80583300039952,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 27047
    },
    {
      "epoch": 10.806232520974831,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 27048
    },
    {
      "epoch": 10.80663204155014,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.051,
      "step": 27049
    },
    {
      "epoch": 10.807031562125449,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 27050
    },
    {
      "epoch": 10.80743108270076,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 27051
    },
    {
      "epoch": 10.807830603276068,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 27052
    },
    {
      "epoch": 10.808230123851379,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 27053
    },
    {
      "epoch": 10.808629644426688,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 27054
    },
    {
      "epoch": 10.809029165001999,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 27055
    },
    {
      "epoch": 10.809428685577307,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 27056
    },
    {
      "epoch": 10.809828206152616,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 27057
    },
    {
      "epoch": 10.810227726727927,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 27058
    },
    {
      "epoch": 10.810627247303236,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 27059
    },
    {
      "epoch": 10.811026767878547,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 27060
    },
    {
      "epoch": 10.811426288453855,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 27061
    },
    {
      "epoch": 10.811825809029164,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 27062
    },
    {
      "epoch": 10.812225329604475,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 27063
    },
    {
      "epoch": 10.812624850179784,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 27064
    },
    {
      "epoch": 10.813024370755095,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 27065
    },
    {
      "epoch": 10.813423891330403,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 27066
    },
    {
      "epoch": 10.813823411905712,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 27067
    },
    {
      "epoch": 10.814222932481023,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 27068
    },
    {
      "epoch": 10.814622453056332,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 27069
    },
    {
      "epoch": 10.815021973631643,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 27070
    },
    {
      "epoch": 10.815421494206952,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 27071
    },
    {
      "epoch": 10.815821014782262,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 27072
    },
    {
      "epoch": 10.816220535357571,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 27073
    },
    {
      "epoch": 10.81662005593288,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 27074
    },
    {
      "epoch": 10.81701957650819,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 27075
    },
    {
      "epoch": 10.8174190970835,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 27076
    },
    {
      "epoch": 10.81781861765881,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 27077
    },
    {
      "epoch": 10.81821813823412,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 27078
    },
    {
      "epoch": 10.818617658809428,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 27079
    },
    {
      "epoch": 10.819017179384739,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 27080
    },
    {
      "epoch": 10.819416699960048,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 27081
    },
    {
      "epoch": 10.819816220535358,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 27082
    },
    {
      "epoch": 10.820215741110667,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 27083
    },
    {
      "epoch": 10.820615261685976,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 27084
    },
    {
      "epoch": 10.821014782261287,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 27085
    },
    {
      "epoch": 10.821414302836596,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 27086
    },
    {
      "epoch": 10.821813823411906,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 27087
    },
    {
      "epoch": 10.822213343987215,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 27088
    },
    {
      "epoch": 10.822612864562524,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 27089
    },
    {
      "epoch": 10.823012385137835,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 27090
    },
    {
      "epoch": 10.823411905713144,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 27091
    },
    {
      "epoch": 10.823811426288454,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 27092
    },
    {
      "epoch": 10.824210946863763,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 27093
    },
    {
      "epoch": 10.824610467439072,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 27094
    },
    {
      "epoch": 10.825009988014383,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 27095
    },
    {
      "epoch": 10.825409508589692,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 27096
    },
    {
      "epoch": 10.825809029165002,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 27097
    },
    {
      "epoch": 10.826208549740311,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 27098
    },
    {
      "epoch": 10.826608070315622,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 27099
    },
    {
      "epoch": 10.827007590890931,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 27100
    },
    {
      "epoch": 10.82740711146624,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 27101
    },
    {
      "epoch": 10.82780663204155,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 27102
    },
    {
      "epoch": 10.82820615261686,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 27103
    },
    {
      "epoch": 10.82860567319217,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 27104
    },
    {
      "epoch": 10.829005193767479,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 27105
    },
    {
      "epoch": 10.829404714342788,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 27106
    },
    {
      "epoch": 10.829804234918099,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 27107
    },
    {
      "epoch": 10.830203755493407,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 27108
    },
    {
      "epoch": 10.830603276068718,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 27109
    },
    {
      "epoch": 10.831002796644027,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 27110
    },
    {
      "epoch": 10.831402317219336,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 27111
    },
    {
      "epoch": 10.831801837794647,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 27112
    },
    {
      "epoch": 10.832201358369955,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 27113
    },
    {
      "epoch": 10.832600878945266,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 27114
    },
    {
      "epoch": 10.833000399520575,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 27115
    },
    {
      "epoch": 10.833399920095886,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 27116
    },
    {
      "epoch": 10.833799440671195,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 27117
    },
    {
      "epoch": 10.834198961246504,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 27118
    },
    {
      "epoch": 10.834598481821814,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 27119
    },
    {
      "epoch": 10.834998002397123,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 27120
    },
    {
      "epoch": 10.835397522972434,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 27121
    },
    {
      "epoch": 10.835797043547743,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 27122
    },
    {
      "epoch": 10.836196564123052,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 27123
    },
    {
      "epoch": 10.836596084698362,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 27124
    },
    {
      "epoch": 10.836995605273671,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 27125
    },
    {
      "epoch": 10.837395125848982,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 27126
    },
    {
      "epoch": 10.83779464642429,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 27127
    },
    {
      "epoch": 10.8381941669996,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 27128
    },
    {
      "epoch": 10.83859368757491,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 27129
    },
    {
      "epoch": 10.83899320815022,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 27130
    },
    {
      "epoch": 10.83939272872553,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 27131
    },
    {
      "epoch": 10.839792249300839,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 27132
    },
    {
      "epoch": 10.84019176987615,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 27133
    },
    {
      "epoch": 10.840591290451458,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 27134
    },
    {
      "epoch": 10.840990811026767,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 27135
    },
    {
      "epoch": 10.841390331602078,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 27136
    },
    {
      "epoch": 10.841789852177387,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 27137
    },
    {
      "epoch": 10.842189372752697,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 27138
    },
    {
      "epoch": 10.842588893328006,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 27139
    },
    {
      "epoch": 10.842988413903315,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 27140
    },
    {
      "epoch": 10.843387934478626,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 27141
    },
    {
      "epoch": 10.843787455053935,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 27142
    },
    {
      "epoch": 10.844186975629245,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 27143
    },
    {
      "epoch": 10.844586496204554,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 27144
    },
    {
      "epoch": 10.844986016779863,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 27145
    },
    {
      "epoch": 10.845385537355174,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 27146
    },
    {
      "epoch": 10.845785057930483,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 27147
    },
    {
      "epoch": 10.846184578505794,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 27148
    },
    {
      "epoch": 10.846584099081102,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 27149
    },
    {
      "epoch": 10.846983619656413,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 27150
    },
    {
      "epoch": 10.847383140231722,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 27151
    },
    {
      "epoch": 10.847782660807031,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 27152
    },
    {
      "epoch": 10.848182181382342,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 27153
    },
    {
      "epoch": 10.84858170195765,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 27154
    },
    {
      "epoch": 10.848981222532961,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 27155
    },
    {
      "epoch": 10.84938074310827,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 27156
    },
    {
      "epoch": 10.849780263683579,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 27157
    },
    {
      "epoch": 10.85017978425889,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 27158
    },
    {
      "epoch": 10.850579304834199,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 27159
    },
    {
      "epoch": 10.85097882540951,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 27160
    },
    {
      "epoch": 10.851378345984818,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 27161
    },
    {
      "epoch": 10.851777866560127,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 27162
    },
    {
      "epoch": 10.852177387135438,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 27163
    },
    {
      "epoch": 10.852576907710747,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 27164
    },
    {
      "epoch": 10.852976428286057,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 27165
    },
    {
      "epoch": 10.853375948861366,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 27166
    },
    {
      "epoch": 10.853775469436677,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 27167
    },
    {
      "epoch": 10.854174990011986,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 27168
    },
    {
      "epoch": 10.854574510587295,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 27169
    },
    {
      "epoch": 10.854974031162605,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 27170
    },
    {
      "epoch": 10.855373551737914,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 27171
    },
    {
      "epoch": 10.855773072313225,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 27172
    },
    {
      "epoch": 10.856172592888534,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 27173
    },
    {
      "epoch": 10.856572113463843,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 27174
    },
    {
      "epoch": 10.856971634039153,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 27175
    },
    {
      "epoch": 10.857371154614462,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 27176
    },
    {
      "epoch": 10.857770675189773,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 27177
    },
    {
      "epoch": 10.858170195765082,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 27178
    },
    {
      "epoch": 10.85856971634039,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 27179
    },
    {
      "epoch": 10.858969236915701,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 27180
    },
    {
      "epoch": 10.85936875749101,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 27181
    },
    {
      "epoch": 10.859768278066321,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 27182
    },
    {
      "epoch": 10.86016779864163,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 27183
    },
    {
      "epoch": 10.86056731921694,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 27184
    },
    {
      "epoch": 10.86096683979225,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 27185
    },
    {
      "epoch": 10.861366360367558,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0492,
      "step": 27186
    },
    {
      "epoch": 10.861765880942869,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 27187
    },
    {
      "epoch": 10.862165401518178,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 27188
    },
    {
      "epoch": 10.862564922093489,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 27189
    },
    {
      "epoch": 10.862964442668797,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 27190
    },
    {
      "epoch": 10.863363963244106,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 27191
    },
    {
      "epoch": 10.863763483819417,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 27192
    },
    {
      "epoch": 10.864163004394726,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 27193
    },
    {
      "epoch": 10.864562524970037,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 27194
    },
    {
      "epoch": 10.864962045545346,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 27195
    },
    {
      "epoch": 10.865361566120654,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 27196
    },
    {
      "epoch": 10.865761086695965,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 27197
    },
    {
      "epoch": 10.866160607271274,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 27198
    },
    {
      "epoch": 10.866560127846585,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 27199
    },
    {
      "epoch": 10.866959648421894,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 27200
    },
    {
      "epoch": 10.867359168997204,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 27201
    },
    {
      "epoch": 10.867758689572513,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 27202
    },
    {
      "epoch": 10.868158210147822,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 27203
    },
    {
      "epoch": 10.868557730723133,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 27204
    },
    {
      "epoch": 10.868957251298442,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 27205
    },
    {
      "epoch": 10.869356771873752,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 27206
    },
    {
      "epoch": 10.869756292449061,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 27207
    },
    {
      "epoch": 10.87015581302437,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 27208
    },
    {
      "epoch": 10.87055533359968,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 27209
    },
    {
      "epoch": 10.87095485417499,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 27210
    },
    {
      "epoch": 10.8713543747503,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 27211
    },
    {
      "epoch": 10.87175389532561,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 27212
    },
    {
      "epoch": 10.872153415900918,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 27213
    },
    {
      "epoch": 10.872552936476229,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 27214
    },
    {
      "epoch": 10.872952457051538,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 27215
    },
    {
      "epoch": 10.873351977626848,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 27216
    },
    {
      "epoch": 10.873751498202157,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 27217
    },
    {
      "epoch": 10.874151018777468,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 27218
    },
    {
      "epoch": 10.874550539352777,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 27219
    },
    {
      "epoch": 10.874950059928086,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 27220
    },
    {
      "epoch": 10.875349580503396,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 27221
    },
    {
      "epoch": 10.875749101078705,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 27222
    },
    {
      "epoch": 10.876148621654016,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 27223
    },
    {
      "epoch": 10.876548142229325,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 27224
    },
    {
      "epoch": 10.876947662804634,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 27225
    },
    {
      "epoch": 10.877347183379944,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 27226
    },
    {
      "epoch": 10.877746703955253,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 27227
    },
    {
      "epoch": 10.878146224530564,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 27228
    },
    {
      "epoch": 10.878545745105873,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 27229
    },
    {
      "epoch": 10.878945265681182,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 27230
    },
    {
      "epoch": 10.879344786256492,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 27231
    },
    {
      "epoch": 10.879744306831801,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 27232
    },
    {
      "epoch": 10.880143827407112,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 27233
    },
    {
      "epoch": 10.880543347982421,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 27234
    },
    {
      "epoch": 10.880942868557732,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 27235
    },
    {
      "epoch": 10.88134238913304,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 27236
    },
    {
      "epoch": 10.88174190970835,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 27237
    },
    {
      "epoch": 10.88214143028366,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 27238
    },
    {
      "epoch": 10.882540950858969,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 27239
    },
    {
      "epoch": 10.88294047143428,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 27240
    },
    {
      "epoch": 10.883339992009589,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 27241
    },
    {
      "epoch": 10.883739512584897,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 27242
    },
    {
      "epoch": 10.884139033160208,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 27243
    },
    {
      "epoch": 10.884538553735517,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 27244
    },
    {
      "epoch": 10.884938074310828,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 27245
    },
    {
      "epoch": 10.885337594886137,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 27246
    },
    {
      "epoch": 10.885737115461446,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 27247
    },
    {
      "epoch": 10.886136636036756,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 27248
    },
    {
      "epoch": 10.886536156612065,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 27249
    },
    {
      "epoch": 10.886935677187376,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 27250
    },
    {
      "epoch": 10.887335197762685,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 27251
    },
    {
      "epoch": 10.887734718337995,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 27252
    },
    {
      "epoch": 10.888134238913304,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 27253
    },
    {
      "epoch": 10.888533759488613,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 27254
    },
    {
      "epoch": 10.888933280063924,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 27255
    },
    {
      "epoch": 10.889332800639233,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 27256
    },
    {
      "epoch": 10.889732321214543,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 27257
    },
    {
      "epoch": 10.890131841789852,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 27258
    },
    {
      "epoch": 10.890531362365161,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 27259
    },
    {
      "epoch": 10.890930882940472,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 27260
    },
    {
      "epoch": 10.89133040351578,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 27261
    },
    {
      "epoch": 10.891729924091091,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 27262
    },
    {
      "epoch": 10.8921294446664,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 27263
    },
    {
      "epoch": 10.89252896524171,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 27264
    },
    {
      "epoch": 10.89292848581702,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 27265
    },
    {
      "epoch": 10.893328006392329,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 27266
    },
    {
      "epoch": 10.89372752696764,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 27267
    },
    {
      "epoch": 10.894127047542948,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 27268
    },
    {
      "epoch": 10.894526568118257,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 27269
    },
    {
      "epoch": 10.894926088693568,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 27270
    },
    {
      "epoch": 10.895325609268877,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 27271
    },
    {
      "epoch": 10.895725129844188,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 27272
    },
    {
      "epoch": 10.896124650419496,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 27273
    },
    {
      "epoch": 10.896524170994805,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 27274
    },
    {
      "epoch": 10.896923691570116,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 27275
    },
    {
      "epoch": 10.897323212145425,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 27276
    },
    {
      "epoch": 10.897722732720736,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 27277
    },
    {
      "epoch": 10.898122253296044,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 27278
    },
    {
      "epoch": 10.898521773871355,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 27279
    },
    {
      "epoch": 10.898921294446664,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 27280
    },
    {
      "epoch": 10.899320815021973,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0498,
      "step": 27281
    },
    {
      "epoch": 10.899720335597284,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 27282
    },
    {
      "epoch": 10.900119856172592,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 27283
    },
    {
      "epoch": 10.900519376747903,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 27284
    },
    {
      "epoch": 10.900918897323212,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 27285
    },
    {
      "epoch": 10.901318417898521,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 27286
    },
    {
      "epoch": 10.901717938473832,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 27287
    },
    {
      "epoch": 10.90211745904914,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 27288
    },
    {
      "epoch": 10.902516979624451,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 27289
    },
    {
      "epoch": 10.90291650019976,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 27290
    },
    {
      "epoch": 10.903316020775069,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 27291
    },
    {
      "epoch": 10.90371554135038,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 27292
    },
    {
      "epoch": 10.904115061925689,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 27293
    },
    {
      "epoch": 10.904514582501,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 27294
    },
    {
      "epoch": 10.904914103076308,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 27295
    },
    {
      "epoch": 10.905313623651619,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 27296
    },
    {
      "epoch": 10.905713144226928,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 27297
    },
    {
      "epoch": 10.906112664802237,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 27298
    },
    {
      "epoch": 10.906512185377547,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 27299
    },
    {
      "epoch": 10.906911705952856,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 27300
    },
    {
      "epoch": 10.907311226528167,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 27301
    },
    {
      "epoch": 10.907710747103476,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 27302
    },
    {
      "epoch": 10.908110267678785,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 27303
    },
    {
      "epoch": 10.908509788254095,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 27304
    },
    {
      "epoch": 10.908909308829404,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 27305
    },
    {
      "epoch": 10.909308829404715,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 27306
    },
    {
      "epoch": 10.909708349980024,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 27307
    },
    {
      "epoch": 10.910107870555333,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 27308
    },
    {
      "epoch": 10.910507391130643,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 27309
    },
    {
      "epoch": 10.910906911705952,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 27310
    },
    {
      "epoch": 10.911306432281263,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 27311
    },
    {
      "epoch": 10.911705952856572,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 27312
    },
    {
      "epoch": 10.912105473431883,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 27313
    },
    {
      "epoch": 10.912504994007191,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 27314
    },
    {
      "epoch": 10.9129045145825,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 27315
    },
    {
      "epoch": 10.913304035157811,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 27316
    },
    {
      "epoch": 10.91370355573312,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 27317
    },
    {
      "epoch": 10.91410307630843,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 27318
    },
    {
      "epoch": 10.91450259688374,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 27319
    },
    {
      "epoch": 10.914902117459048,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 27320
    },
    {
      "epoch": 10.915301638034359,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 27321
    },
    {
      "epoch": 10.915701158609668,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 27322
    },
    {
      "epoch": 10.916100679184979,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 27323
    },
    {
      "epoch": 10.916500199760288,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 27324
    },
    {
      "epoch": 10.916899720335596,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 27325
    },
    {
      "epoch": 10.917299240910907,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 27326
    },
    {
      "epoch": 10.917698761486216,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 27327
    },
    {
      "epoch": 10.918098282061527,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 27328
    },
    {
      "epoch": 10.918497802636836,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 27329
    },
    {
      "epoch": 10.918897323212146,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 27330
    },
    {
      "epoch": 10.919296843787455,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 27331
    },
    {
      "epoch": 10.919696364362764,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 27332
    },
    {
      "epoch": 10.920095884938075,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 27333
    },
    {
      "epoch": 10.920495405513384,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 27334
    },
    {
      "epoch": 10.920894926088694,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 27335
    },
    {
      "epoch": 10.921294446664003,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 27336
    },
    {
      "epoch": 10.921693967239312,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 27337
    },
    {
      "epoch": 10.922093487814623,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 27338
    },
    {
      "epoch": 10.922493008389932,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 27339
    },
    {
      "epoch": 10.922892528965242,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 27340
    },
    {
      "epoch": 10.923292049540551,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 27341
    },
    {
      "epoch": 10.92369157011586,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 27342
    },
    {
      "epoch": 10.92409109069117,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 27343
    },
    {
      "epoch": 10.92449061126648,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 27344
    },
    {
      "epoch": 10.92489013184179,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 27345
    },
    {
      "epoch": 10.9252896524171,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 27346
    },
    {
      "epoch": 10.92568917299241,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 27347
    },
    {
      "epoch": 10.926088693567719,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 27348
    },
    {
      "epoch": 10.926488214143028,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 27349
    },
    {
      "epoch": 10.926887734718338,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 27350
    },
    {
      "epoch": 10.927287255293647,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 27351
    },
    {
      "epoch": 10.927686775868958,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 27352
    },
    {
      "epoch": 10.928086296444267,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 27353
    },
    {
      "epoch": 10.928485817019576,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 27354
    },
    {
      "epoch": 10.928885337594886,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 27355
    },
    {
      "epoch": 10.929284858170195,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 27356
    },
    {
      "epoch": 10.929684378745506,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 27357
    },
    {
      "epoch": 10.930083899320815,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 27358
    },
    {
      "epoch": 10.930483419896124,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 27359
    },
    {
      "epoch": 10.930882940471434,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 27360
    },
    {
      "epoch": 10.931282461046743,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 27361
    },
    {
      "epoch": 10.931681981622054,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 27362
    },
    {
      "epoch": 10.932081502197363,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 27363
    },
    {
      "epoch": 10.932481022772674,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 27364
    },
    {
      "epoch": 10.932880543347983,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 27365
    },
    {
      "epoch": 10.933280063923291,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 27366
    },
    {
      "epoch": 10.933679584498602,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 27367
    },
    {
      "epoch": 10.934079105073911,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 27368
    },
    {
      "epoch": 10.934478625649222,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 27369
    },
    {
      "epoch": 10.93487814622453,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 27370
    },
    {
      "epoch": 10.93527766679984,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 27371
    },
    {
      "epoch": 10.93567718737515,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 27372
    },
    {
      "epoch": 10.936076707950459,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 27373
    },
    {
      "epoch": 10.93647622852577,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 27374
    },
    {
      "epoch": 10.936875749101079,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 27375
    },
    {
      "epoch": 10.937275269676388,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 27376
    },
    {
      "epoch": 10.937674790251698,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 27377
    },
    {
      "epoch": 10.938074310827007,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 27378
    },
    {
      "epoch": 10.938473831402318,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 27379
    },
    {
      "epoch": 10.938873351977627,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 27380
    },
    {
      "epoch": 10.939272872552937,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 27381
    },
    {
      "epoch": 10.939672393128246,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 27382
    },
    {
      "epoch": 10.940071913703555,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 27383
    },
    {
      "epoch": 10.940471434278866,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 27384
    },
    {
      "epoch": 10.940870954854175,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 27385
    },
    {
      "epoch": 10.941270475429485,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 27386
    },
    {
      "epoch": 10.941669996004794,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 27387
    },
    {
      "epoch": 10.942069516580103,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 27388
    },
    {
      "epoch": 10.942469037155414,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 27389
    },
    {
      "epoch": 10.942868557730723,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 27390
    },
    {
      "epoch": 10.943268078306033,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 27391
    },
    {
      "epoch": 10.943667598881342,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 27392
    },
    {
      "epoch": 10.944067119456651,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 27393
    },
    {
      "epoch": 10.944466640031962,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 27394
    },
    {
      "epoch": 10.94486616060727,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 27395
    },
    {
      "epoch": 10.945265681182581,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 27396
    },
    {
      "epoch": 10.94566520175789,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 27397
    },
    {
      "epoch": 10.946064722333201,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 27398
    },
    {
      "epoch": 10.94646424290851,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 27399
    },
    {
      "epoch": 10.946863763483819,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 27400
    },
    {
      "epoch": 10.94726328405913,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 27401
    },
    {
      "epoch": 10.947662804634438,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 27402
    },
    {
      "epoch": 10.948062325209749,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 27403
    },
    {
      "epoch": 10.948461845785058,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 27404
    },
    {
      "epoch": 10.948861366360367,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 27405
    },
    {
      "epoch": 10.949260886935678,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 27406
    },
    {
      "epoch": 10.949660407510986,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 27407
    },
    {
      "epoch": 10.950059928086297,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 27408
    },
    {
      "epoch": 10.950459448661606,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 27409
    },
    {
      "epoch": 10.950858969236915,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 27410
    },
    {
      "epoch": 10.951258489812226,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 27411
    },
    {
      "epoch": 10.951658010387535,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 27412
    },
    {
      "epoch": 10.952057530962845,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 27413
    },
    {
      "epoch": 10.952457051538154,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 27414
    },
    {
      "epoch": 10.952856572113465,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 27415
    },
    {
      "epoch": 10.953256092688774,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 27416
    },
    {
      "epoch": 10.953655613264083,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 27417
    },
    {
      "epoch": 10.954055133839393,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 27418
    },
    {
      "epoch": 10.954454654414702,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 27419
    },
    {
      "epoch": 10.954854174990013,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 27420
    },
    {
      "epoch": 10.955253695565322,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 27421
    },
    {
      "epoch": 10.95565321614063,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 27422
    },
    {
      "epoch": 10.956052736715941,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 27423
    },
    {
      "epoch": 10.95645225729125,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 27424
    },
    {
      "epoch": 10.95685177786656,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 27425
    },
    {
      "epoch": 10.95725129844187,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 27426
    },
    {
      "epoch": 10.957650819017179,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 27427
    },
    {
      "epoch": 10.95805033959249,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 27428
    },
    {
      "epoch": 10.958449860167798,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 27429
    },
    {
      "epoch": 10.958849380743109,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 27430
    },
    {
      "epoch": 10.959248901318418,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 27431
    },
    {
      "epoch": 10.959648421893728,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 27432
    },
    {
      "epoch": 10.960047942469037,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 27433
    },
    {
      "epoch": 10.960447463044346,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 27434
    },
    {
      "epoch": 10.960846983619657,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 27435
    },
    {
      "epoch": 10.961246504194966,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 27436
    },
    {
      "epoch": 10.961646024770276,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 27437
    },
    {
      "epoch": 10.962045545345585,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 27438
    },
    {
      "epoch": 10.962445065920894,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 27439
    },
    {
      "epoch": 10.962844586496205,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 27440
    },
    {
      "epoch": 10.963244107071514,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 27441
    },
    {
      "epoch": 10.963643627646825,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 27442
    },
    {
      "epoch": 10.964043148222133,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 27443
    },
    {
      "epoch": 10.964442668797442,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 27444
    },
    {
      "epoch": 10.964842189372753,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 27445
    },
    {
      "epoch": 10.965241709948062,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 27446
    },
    {
      "epoch": 10.965641230523373,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 27447
    },
    {
      "epoch": 10.966040751098681,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 27448
    },
    {
      "epoch": 10.96644027167399,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 27449
    },
    {
      "epoch": 10.966839792249301,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 27450
    },
    {
      "epoch": 10.96723931282461,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 27451
    },
    {
      "epoch": 10.96763883339992,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 27452
    },
    {
      "epoch": 10.96803835397523,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 27453
    },
    {
      "epoch": 10.968437874550538,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 27454
    },
    {
      "epoch": 10.968837395125849,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 27455
    },
    {
      "epoch": 10.969236915701158,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 27456
    },
    {
      "epoch": 10.969636436276469,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 27457
    },
    {
      "epoch": 10.970035956851778,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 27458
    },
    {
      "epoch": 10.970435477427088,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 27459
    },
    {
      "epoch": 10.970834998002397,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 27460
    },
    {
      "epoch": 10.971234518577706,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 27461
    },
    {
      "epoch": 10.971634039153017,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 27462
    },
    {
      "epoch": 10.972033559728326,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 27463
    },
    {
      "epoch": 10.972433080303636,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 27464
    },
    {
      "epoch": 10.972832600878945,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 27465
    },
    {
      "epoch": 10.973232121454254,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 27466
    },
    {
      "epoch": 10.973631642029565,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 27467
    },
    {
      "epoch": 10.974031162604874,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 27468
    },
    {
      "epoch": 10.974430683180184,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 27469
    },
    {
      "epoch": 10.974830203755493,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 27470
    },
    {
      "epoch": 10.975229724330802,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 27471
    },
    {
      "epoch": 10.975629244906113,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 27472
    },
    {
      "epoch": 10.976028765481422,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 27473
    },
    {
      "epoch": 10.976428286056732,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 27474
    },
    {
      "epoch": 10.976827806632041,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 27475
    },
    {
      "epoch": 10.977227327207352,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 27476
    },
    {
      "epoch": 10.97762684778266,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 27477
    },
    {
      "epoch": 10.97802636835797,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 27478
    },
    {
      "epoch": 10.97842588893328,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 27479
    },
    {
      "epoch": 10.97882540950859,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 27480
    },
    {
      "epoch": 10.9792249300839,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 27481
    },
    {
      "epoch": 10.979624450659209,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 27482
    },
    {
      "epoch": 10.980023971234518,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 27483
    },
    {
      "epoch": 10.980423491809828,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 27484
    },
    {
      "epoch": 10.980823012385137,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 27485
    },
    {
      "epoch": 10.981222532960448,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 27486
    },
    {
      "epoch": 10.981622053535757,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 27487
    },
    {
      "epoch": 10.982021574111066,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 27488
    },
    {
      "epoch": 10.982421094686377,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 27489
    },
    {
      "epoch": 10.982820615261685,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 27490
    },
    {
      "epoch": 10.983220135836996,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 27491
    },
    {
      "epoch": 10.983619656412305,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 27492
    },
    {
      "epoch": 10.984019176987616,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 27493
    },
    {
      "epoch": 10.984418697562925,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 27494
    },
    {
      "epoch": 10.984818218138233,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 27495
    },
    {
      "epoch": 10.985217738713544,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 27496
    },
    {
      "epoch": 10.985617259288853,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 27497
    },
    {
      "epoch": 10.986016779864164,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 27498
    },
    {
      "epoch": 10.986416300439473,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 27499
    },
    {
      "epoch": 10.986815821014781,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 27500
    },
    {
      "epoch": 10.987215341590092,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 27501
    },
    {
      "epoch": 10.987614862165401,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0524,
      "step": 27502
    },
    {
      "epoch": 10.988014382740712,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 27503
    },
    {
      "epoch": 10.98841390331602,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 27504
    },
    {
      "epoch": 10.98881342389133,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 27505
    },
    {
      "epoch": 10.98921294446664,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 27506
    },
    {
      "epoch": 10.989612465041949,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 27507
    },
    {
      "epoch": 10.99001198561726,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 27508
    },
    {
      "epoch": 10.990411506192569,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 27509
    },
    {
      "epoch": 10.99081102676788,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 27510
    },
    {
      "epoch": 10.991210547343188,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 27511
    },
    {
      "epoch": 10.991610067918497,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 27512
    },
    {
      "epoch": 10.992009588493808,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 27513
    },
    {
      "epoch": 10.992409109069117,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 27514
    },
    {
      "epoch": 10.992808629644427,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 27515
    },
    {
      "epoch": 10.993208150219736,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 27516
    },
    {
      "epoch": 10.993607670795045,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 27517
    },
    {
      "epoch": 10.994007191370356,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 27518
    },
    {
      "epoch": 10.994406711945665,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 27519
    },
    {
      "epoch": 10.994806232520975,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 27520
    },
    {
      "epoch": 10.995205753096284,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 27521
    },
    {
      "epoch": 10.995605273671593,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 27522
    },
    {
      "epoch": 10.996004794246904,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 27523
    },
    {
      "epoch": 10.996404314822213,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 27524
    },
    {
      "epoch": 10.996803835397523,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 27525
    },
    {
      "epoch": 10.997203355972832,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 27526
    },
    {
      "epoch": 10.997602876548143,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 27527
    },
    {
      "epoch": 10.998002397123452,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 27528
    },
    {
      "epoch": 10.99840191769876,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 27529
    },
    {
      "epoch": 10.998801438274072,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 27530
    },
    {
      "epoch": 10.99920095884938,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 27531
    },
    {
      "epoch": 10.999600479424691,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 27532
    },
    {
      "epoch": 11.0,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 27533
    },
    {
      "epoch": 11.000399520575309,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 27534
    },
    {
      "epoch": 11.00079904115062,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 27535
    },
    {
      "epoch": 11.001198561725928,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 27536
    },
    {
      "epoch": 11.00159808230124,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 27537
    },
    {
      "epoch": 11.001997602876548,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 27538
    },
    {
      "epoch": 11.002397123451857,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 27539
    },
    {
      "epoch": 11.002796644027168,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 27540
    },
    {
      "epoch": 11.003196164602477,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 27541
    },
    {
      "epoch": 11.003595685177787,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 27542
    },
    {
      "epoch": 11.003995205753096,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 27543
    },
    {
      "epoch": 11.004394726328407,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 27544
    },
    {
      "epoch": 11.004794246903716,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 27545
    },
    {
      "epoch": 11.005193767479025,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 27546
    },
    {
      "epoch": 11.005593288054335,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 27547
    },
    {
      "epoch": 11.005992808629644,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 27548
    },
    {
      "epoch": 11.006392329204955,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 27549
    },
    {
      "epoch": 11.006791849780264,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 27550
    },
    {
      "epoch": 11.007191370355573,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 27551
    },
    {
      "epoch": 11.007590890930883,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 27552
    },
    {
      "epoch": 11.007990411506192,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 27553
    },
    {
      "epoch": 11.008389932081503,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 27554
    },
    {
      "epoch": 11.008789452656812,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 27555
    },
    {
      "epoch": 11.00918897323212,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 27556
    },
    {
      "epoch": 11.009588493807431,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 27557
    },
    {
      "epoch": 11.00998801438274,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 27558
    },
    {
      "epoch": 11.010387534958051,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 27559
    },
    {
      "epoch": 11.01078705553336,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 27560
    },
    {
      "epoch": 11.01118657610867,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 27561
    },
    {
      "epoch": 11.01158609668398,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 27562
    },
    {
      "epoch": 11.011985617259288,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 27563
    },
    {
      "epoch": 11.012385137834599,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 27564
    },
    {
      "epoch": 11.012784658409908,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 27565
    },
    {
      "epoch": 11.013184178985219,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 27566
    },
    {
      "epoch": 11.013583699560527,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 27567
    },
    {
      "epoch": 11.013983220135836,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 27568
    },
    {
      "epoch": 11.014382740711147,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 27569
    },
    {
      "epoch": 11.014782261286456,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 27570
    },
    {
      "epoch": 11.015181781861767,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 27571
    },
    {
      "epoch": 11.015581302437075,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 27572
    },
    {
      "epoch": 11.015980823012384,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 27573
    },
    {
      "epoch": 11.016380343587695,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 27574
    },
    {
      "epoch": 11.016779864163004,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 27575
    },
    {
      "epoch": 11.017179384738315,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 27576
    },
    {
      "epoch": 11.017578905313623,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 27577
    },
    {
      "epoch": 11.017978425888934,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 27578
    },
    {
      "epoch": 11.018377946464243,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 27579
    },
    {
      "epoch": 11.018777467039552,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 27580
    },
    {
      "epoch": 11.019176987614863,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 27581
    },
    {
      "epoch": 11.019576508190172,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 27582
    },
    {
      "epoch": 11.019976028765482,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 27583
    },
    {
      "epoch": 11.020375549340791,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 27584
    },
    {
      "epoch": 11.0207750699161,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 27585
    },
    {
      "epoch": 11.02117459049141,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 27586
    },
    {
      "epoch": 11.02157411106672,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 27587
    },
    {
      "epoch": 11.02197363164203,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 27588
    },
    {
      "epoch": 11.02237315221734,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 27589
    },
    {
      "epoch": 11.022772672792648,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 27590
    },
    {
      "epoch": 11.023172193367959,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 27591
    },
    {
      "epoch": 11.023571713943268,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 27592
    },
    {
      "epoch": 11.023971234518578,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 27593
    },
    {
      "epoch": 11.024370755093887,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 27594
    },
    {
      "epoch": 11.024770275669196,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 27595
    },
    {
      "epoch": 11.025169796244507,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 27596
    },
    {
      "epoch": 11.025569316819816,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 27597
    },
    {
      "epoch": 11.025968837395126,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 27598
    },
    {
      "epoch": 11.026368357970435,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 27599
    },
    {
      "epoch": 11.026767878545746,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 27600
    },
    {
      "epoch": 11.027167399121055,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 27601
    },
    {
      "epoch": 11.027566919696364,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 27602
    },
    {
      "epoch": 11.027966440271674,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 27603
    },
    {
      "epoch": 11.028365960846983,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 27604
    },
    {
      "epoch": 11.028765481422294,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 27605
    },
    {
      "epoch": 11.029165001997603,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 27606
    },
    {
      "epoch": 11.029564522572912,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 27607
    },
    {
      "epoch": 11.029964043148222,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 27608
    },
    {
      "epoch": 11.030363563723531,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 27609
    },
    {
      "epoch": 11.030763084298842,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 27610
    },
    {
      "epoch": 11.031162604874151,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 27611
    },
    {
      "epoch": 11.03156212544946,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 27612
    },
    {
      "epoch": 11.03196164602477,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 27613
    },
    {
      "epoch": 11.03236116660008,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 27614
    },
    {
      "epoch": 11.03276068717539,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 27615
    },
    {
      "epoch": 11.033160207750699,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 27616
    },
    {
      "epoch": 11.03355972832601,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 27617
    },
    {
      "epoch": 11.033959248901319,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 27618
    },
    {
      "epoch": 11.034358769476627,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 27619
    },
    {
      "epoch": 11.034758290051938,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 27620
    },
    {
      "epoch": 11.035157810627247,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 27621
    },
    {
      "epoch": 11.035557331202558,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 27622
    },
    {
      "epoch": 11.035956851777867,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 27623
    },
    {
      "epoch": 11.036356372353175,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 27624
    },
    {
      "epoch": 11.036755892928486,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 27625
    },
    {
      "epoch": 11.037155413503795,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 27626
    },
    {
      "epoch": 11.037554934079106,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 27627
    },
    {
      "epoch": 11.037954454654415,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 27628
    },
    {
      "epoch": 11.038353975229724,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 27629
    },
    {
      "epoch": 11.038753495805034,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 27630
    },
    {
      "epoch": 11.039153016380343,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 27631
    },
    {
      "epoch": 11.039552536955654,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 27632
    },
    {
      "epoch": 11.039952057530963,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 27633
    },
    {
      "epoch": 11.040351578106273,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 27634
    },
    {
      "epoch": 11.040751098681582,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 27635
    },
    {
      "epoch": 11.041150619256891,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 27636
    },
    {
      "epoch": 11.041550139832202,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 27637
    },
    {
      "epoch": 11.04194966040751,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 27638
    },
    {
      "epoch": 11.042349180982821,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 27639
    },
    {
      "epoch": 11.04274870155813,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 27640
    },
    {
      "epoch": 11.04314822213344,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 27641
    },
    {
      "epoch": 11.04354774270875,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 27642
    },
    {
      "epoch": 11.043947263284059,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 27643
    },
    {
      "epoch": 11.04434678385937,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 27644
    },
    {
      "epoch": 11.044746304434678,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 27645
    },
    {
      "epoch": 11.045145825009987,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 27646
    },
    {
      "epoch": 11.045545345585298,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 27647
    },
    {
      "epoch": 11.045944866160607,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 27648
    },
    {
      "epoch": 11.046344386735917,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 27649
    },
    {
      "epoch": 11.046743907311226,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 27650
    },
    {
      "epoch": 11.047143427886537,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 27651
    },
    {
      "epoch": 11.047542948461846,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 27652
    },
    {
      "epoch": 11.047942469037155,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 27653
    },
    {
      "epoch": 11.048341989612465,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 27654
    },
    {
      "epoch": 11.048741510187774,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 27655
    },
    {
      "epoch": 11.049141030763085,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 27656
    },
    {
      "epoch": 11.049540551338394,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 27657
    },
    {
      "epoch": 11.049940071913703,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 27658
    },
    {
      "epoch": 11.050339592489014,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0134,
      "step": 27659
    },
    {
      "epoch": 11.050739113064322,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 27660
    },
    {
      "epoch": 11.051138633639633,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 27661
    },
    {
      "epoch": 11.051538154214942,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 27662
    },
    {
      "epoch": 11.051937674790251,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 27663
    },
    {
      "epoch": 11.052337195365562,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 27664
    },
    {
      "epoch": 11.05273671594087,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 27665
    },
    {
      "epoch": 11.053136236516181,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 27666
    },
    {
      "epoch": 11.05353575709149,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 27667
    },
    {
      "epoch": 11.053935277666799,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 27668
    },
    {
      "epoch": 11.05433479824211,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 27669
    },
    {
      "epoch": 11.054734318817419,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 27670
    },
    {
      "epoch": 11.05513383939273,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 27671
    },
    {
      "epoch": 11.055533359968038,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 27672
    },
    {
      "epoch": 11.055932880543349,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 27673
    },
    {
      "epoch": 11.056332401118658,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 27674
    },
    {
      "epoch": 11.056731921693967,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 27675
    },
    {
      "epoch": 11.057131442269277,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 27676
    },
    {
      "epoch": 11.057530962844586,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 27677
    },
    {
      "epoch": 11.057930483419897,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 27678
    },
    {
      "epoch": 11.058330003995206,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 27679
    },
    {
      "epoch": 11.058729524570515,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 27680
    },
    {
      "epoch": 11.059129045145825,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 27681
    },
    {
      "epoch": 11.059528565721134,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 27682
    },
    {
      "epoch": 11.059928086296445,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 27683
    },
    {
      "epoch": 11.060327606871754,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 27684
    },
    {
      "epoch": 11.060727127447063,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 27685
    },
    {
      "epoch": 11.061126648022373,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 27686
    },
    {
      "epoch": 11.061526168597682,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 27687
    },
    {
      "epoch": 11.061925689172993,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 27688
    },
    {
      "epoch": 11.062325209748302,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 27689
    },
    {
      "epoch": 11.062724730323612,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 27690
    },
    {
      "epoch": 11.063124250898921,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 27691
    },
    {
      "epoch": 11.06352377147423,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 27692
    },
    {
      "epoch": 11.063923292049541,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 27693
    },
    {
      "epoch": 11.06432281262485,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 27694
    },
    {
      "epoch": 11.06472233320016,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 27695
    },
    {
      "epoch": 11.06512185377547,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 27696
    },
    {
      "epoch": 11.065521374350778,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 27697
    },
    {
      "epoch": 11.065920894926089,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 27698
    },
    {
      "epoch": 11.066320415501398,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 27699
    },
    {
      "epoch": 11.066719936076709,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 27700
    },
    {
      "epoch": 11.067119456652017,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 27701
    },
    {
      "epoch": 11.067518977227326,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 27702
    },
    {
      "epoch": 11.067918497802637,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 27703
    },
    {
      "epoch": 11.068318018377946,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 27704
    },
    {
      "epoch": 11.068717538953257,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 27705
    },
    {
      "epoch": 11.069117059528566,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 27706
    },
    {
      "epoch": 11.069516580103876,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 27707
    },
    {
      "epoch": 11.069916100679185,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 27708
    },
    {
      "epoch": 11.070315621254494,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 27709
    },
    {
      "epoch": 11.070715141829805,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 27710
    },
    {
      "epoch": 11.071114662405114,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 27711
    },
    {
      "epoch": 11.071514182980424,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 27712
    },
    {
      "epoch": 11.071913703555733,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 27713
    },
    {
      "epoch": 11.072313224131042,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 27714
    },
    {
      "epoch": 11.072712744706353,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 27715
    },
    {
      "epoch": 11.073112265281662,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 27716
    },
    {
      "epoch": 11.073511785856972,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 27717
    },
    {
      "epoch": 11.073911306432281,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 27718
    },
    {
      "epoch": 11.07431082700759,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 27719
    },
    {
      "epoch": 11.0747103475829,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 27720
    },
    {
      "epoch": 11.07510986815821,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 27721
    },
    {
      "epoch": 11.07550938873352,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 27722
    },
    {
      "epoch": 11.07590890930883,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 27723
    },
    {
      "epoch": 11.07630842988414,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 27724
    },
    {
      "epoch": 11.076707950459449,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 27725
    },
    {
      "epoch": 11.077107471034758,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 27726
    },
    {
      "epoch": 11.077506991610068,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 27727
    },
    {
      "epoch": 11.077906512185377,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 27728
    },
    {
      "epoch": 11.078306032760688,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 27729
    },
    {
      "epoch": 11.078705553335997,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 27730
    },
    {
      "epoch": 11.079105073911306,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 27731
    },
    {
      "epoch": 11.079504594486616,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 27732
    },
    {
      "epoch": 11.079904115061925,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 27733
    },
    {
      "epoch": 11.080303635637236,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 27734
    },
    {
      "epoch": 11.080703156212545,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 27735
    },
    {
      "epoch": 11.081102676787854,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 27736
    },
    {
      "epoch": 11.081502197363164,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 27737
    },
    {
      "epoch": 11.081901717938473,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 27738
    },
    {
      "epoch": 11.082301238513784,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 27739
    },
    {
      "epoch": 11.082700759089093,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 27740
    },
    {
      "epoch": 11.083100279664404,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 27741
    },
    {
      "epoch": 11.083499800239712,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 27742
    },
    {
      "epoch": 11.083899320815021,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 27743
    },
    {
      "epoch": 11.084298841390332,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 27744
    },
    {
      "epoch": 11.084698361965641,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 27745
    },
    {
      "epoch": 11.085097882540952,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 27746
    },
    {
      "epoch": 11.08549740311626,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 27747
    },
    {
      "epoch": 11.08589692369157,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 27748
    },
    {
      "epoch": 11.08629644426688,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 27749
    },
    {
      "epoch": 11.086695964842189,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 27750
    },
    {
      "epoch": 11.0870954854175,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 27751
    },
    {
      "epoch": 11.087495005992809,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 27752
    },
    {
      "epoch": 11.087894526568117,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 27753
    },
    {
      "epoch": 11.088294047143428,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 27754
    },
    {
      "epoch": 11.088693567718737,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 27755
    },
    {
      "epoch": 11.089093088294048,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 27756
    },
    {
      "epoch": 11.089492608869357,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 27757
    },
    {
      "epoch": 11.089892129444667,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 27758
    },
    {
      "epoch": 11.090291650019976,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 27759
    },
    {
      "epoch": 11.090691170595285,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 27760
    },
    {
      "epoch": 11.091090691170596,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 27761
    },
    {
      "epoch": 11.091490211745905,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 27762
    },
    {
      "epoch": 11.091889732321215,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 27763
    },
    {
      "epoch": 11.092289252896524,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 27764
    },
    {
      "epoch": 11.092688773471833,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 27765
    },
    {
      "epoch": 11.093088294047144,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 27766
    },
    {
      "epoch": 11.093487814622453,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 27767
    },
    {
      "epoch": 11.093887335197763,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 27768
    },
    {
      "epoch": 11.094286855773072,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0512,
      "step": 27769
    },
    {
      "epoch": 11.094686376348381,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 27770
    },
    {
      "epoch": 11.095085896923692,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 27771
    },
    {
      "epoch": 11.095485417499,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 27772
    },
    {
      "epoch": 11.095884938074311,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 27773
    },
    {
      "epoch": 11.09628445864962,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 27774
    },
    {
      "epoch": 11.09668397922493,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 27775
    },
    {
      "epoch": 11.09708349980024,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 27776
    },
    {
      "epoch": 11.097483020375549,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 27777
    },
    {
      "epoch": 11.09788254095086,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 27778
    },
    {
      "epoch": 11.098282061526168,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 27779
    },
    {
      "epoch": 11.098681582101479,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 27780
    },
    {
      "epoch": 11.099081102676788,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 27781
    },
    {
      "epoch": 11.099480623252097,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 27782
    },
    {
      "epoch": 11.099880143827408,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 27783
    },
    {
      "epoch": 11.100279664402716,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 27784
    },
    {
      "epoch": 11.100679184978027,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 27785
    },
    {
      "epoch": 11.101078705553336,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 27786
    },
    {
      "epoch": 11.101478226128645,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 27787
    },
    {
      "epoch": 11.101877746703956,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 27788
    },
    {
      "epoch": 11.102277267279264,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 27789
    },
    {
      "epoch": 11.102676787854575,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 27790
    },
    {
      "epoch": 11.103076308429884,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 27791
    },
    {
      "epoch": 11.103475829005193,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 27792
    },
    {
      "epoch": 11.103875349580504,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 27793
    },
    {
      "epoch": 11.104274870155812,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 27794
    },
    {
      "epoch": 11.104674390731123,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 27795
    },
    {
      "epoch": 11.105073911306432,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 27796
    },
    {
      "epoch": 11.105473431881743,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 27797
    },
    {
      "epoch": 11.105872952457052,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 27798
    },
    {
      "epoch": 11.10627247303236,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 27799
    },
    {
      "epoch": 11.106671993607671,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 27800
    },
    {
      "epoch": 11.10707151418298,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 27801
    },
    {
      "epoch": 11.10747103475829,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 27802
    },
    {
      "epoch": 11.1078705553336,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 27803
    },
    {
      "epoch": 11.108270075908909,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 27804
    },
    {
      "epoch": 11.10866959648422,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 27805
    },
    {
      "epoch": 11.109069117059528,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 27806
    },
    {
      "epoch": 11.109468637634839,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 27807
    },
    {
      "epoch": 11.109868158210148,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 27808
    },
    {
      "epoch": 11.110267678785457,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 27809
    },
    {
      "epoch": 11.110667199360767,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 27810
    },
    {
      "epoch": 11.111066719936076,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 27811
    },
    {
      "epoch": 11.111466240511387,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 27812
    },
    {
      "epoch": 11.111865761086696,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 27813
    },
    {
      "epoch": 11.112265281662006,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 27814
    },
    {
      "epoch": 11.112664802237315,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 27815
    },
    {
      "epoch": 11.113064322812624,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 27816
    },
    {
      "epoch": 11.113463843387935,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 27817
    },
    {
      "epoch": 11.113863363963244,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 27818
    },
    {
      "epoch": 11.114262884538554,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 27819
    },
    {
      "epoch": 11.114662405113863,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 27820
    },
    {
      "epoch": 11.115061925689172,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 27821
    },
    {
      "epoch": 11.115461446264483,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 27822
    },
    {
      "epoch": 11.115860966839792,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 27823
    },
    {
      "epoch": 11.116260487415103,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 27824
    },
    {
      "epoch": 11.116660007990411,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 27825
    },
    {
      "epoch": 11.11705952856572,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 27826
    },
    {
      "epoch": 11.117459049141031,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 27827
    },
    {
      "epoch": 11.11785856971634,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 27828
    },
    {
      "epoch": 11.11825809029165,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 27829
    },
    {
      "epoch": 11.11865761086696,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 27830
    },
    {
      "epoch": 11.11905713144227,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 27831
    },
    {
      "epoch": 11.119456652017579,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 27832
    },
    {
      "epoch": 11.119856172592888,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 27833
    },
    {
      "epoch": 11.120255693168199,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 27834
    },
    {
      "epoch": 11.120655213743508,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 27835
    },
    {
      "epoch": 11.121054734318818,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 27836
    },
    {
      "epoch": 11.121454254894127,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 27837
    },
    {
      "epoch": 11.121853775469436,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 27838
    },
    {
      "epoch": 11.122253296044747,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.012,
      "step": 27839
    },
    {
      "epoch": 11.122652816620056,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 27840
    },
    {
      "epoch": 11.123052337195366,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 27841
    },
    {
      "epoch": 11.123451857770675,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 27842
    },
    {
      "epoch": 11.123851378345984,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 27843
    },
    {
      "epoch": 11.124250898921295,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 27844
    },
    {
      "epoch": 11.124650419496604,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 27845
    },
    {
      "epoch": 11.125049940071914,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 27846
    },
    {
      "epoch": 11.125449460647223,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 27847
    },
    {
      "epoch": 11.125848981222532,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 27848
    },
    {
      "epoch": 11.126248501797843,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 27849
    },
    {
      "epoch": 11.126648022373152,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 27850
    },
    {
      "epoch": 11.127047542948462,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 27851
    },
    {
      "epoch": 11.127447063523771,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 27852
    },
    {
      "epoch": 11.127846584099082,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 27853
    },
    {
      "epoch": 11.12824610467439,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 27854
    },
    {
      "epoch": 11.1286456252497,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 27855
    },
    {
      "epoch": 11.12904514582501,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 27856
    },
    {
      "epoch": 11.12944466640032,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0527,
      "step": 27857
    },
    {
      "epoch": 11.12984418697563,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 27858
    },
    {
      "epoch": 11.130243707550939,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 27859
    },
    {
      "epoch": 11.130643228126248,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 27860
    },
    {
      "epoch": 11.131042748701558,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 27861
    },
    {
      "epoch": 11.131442269276867,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 27862
    },
    {
      "epoch": 11.131841789852178,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 27863
    },
    {
      "epoch": 11.132241310427487,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 27864
    },
    {
      "epoch": 11.132640831002796,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 27865
    },
    {
      "epoch": 11.133040351578106,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 27866
    },
    {
      "epoch": 11.133439872153415,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 27867
    },
    {
      "epoch": 11.133839392728726,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 27868
    },
    {
      "epoch": 11.134238913304035,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 27869
    },
    {
      "epoch": 11.134638433879346,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 27870
    },
    {
      "epoch": 11.135037954454654,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 27871
    },
    {
      "epoch": 11.135437475029963,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 27872
    },
    {
      "epoch": 11.135836995605274,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 27873
    },
    {
      "epoch": 11.136236516180583,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 27874
    },
    {
      "epoch": 11.136636036755894,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 27875
    },
    {
      "epoch": 11.137035557331203,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 27876
    },
    {
      "epoch": 11.137435077906511,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 27877
    },
    {
      "epoch": 11.137834598481822,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 27878
    },
    {
      "epoch": 11.138234119057131,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 27879
    },
    {
      "epoch": 11.138633639632442,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 27880
    },
    {
      "epoch": 11.13903316020775,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 27881
    },
    {
      "epoch": 11.13943268078306,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 27882
    },
    {
      "epoch": 11.13983220135837,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 27883
    },
    {
      "epoch": 11.140231721933679,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 27884
    },
    {
      "epoch": 11.14063124250899,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 27885
    },
    {
      "epoch": 11.141030763084299,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 27886
    },
    {
      "epoch": 11.14143028365961,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 27887
    },
    {
      "epoch": 11.141829804234918,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 27888
    },
    {
      "epoch": 11.142229324810227,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 27889
    },
    {
      "epoch": 11.142628845385538,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 27890
    },
    {
      "epoch": 11.143028365960847,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 27891
    },
    {
      "epoch": 11.143427886536157,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 27892
    },
    {
      "epoch": 11.143827407111466,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0481,
      "step": 27893
    },
    {
      "epoch": 11.144226927686775,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 27894
    },
    {
      "epoch": 11.144626448262086,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 27895
    },
    {
      "epoch": 11.145025968837395,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 27896
    },
    {
      "epoch": 11.145425489412705,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 27897
    },
    {
      "epoch": 11.145825009988014,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 27898
    },
    {
      "epoch": 11.146224530563323,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 27899
    },
    {
      "epoch": 11.146624051138634,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 27900
    },
    {
      "epoch": 11.147023571713943,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 27901
    },
    {
      "epoch": 11.147423092289253,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 27902
    },
    {
      "epoch": 11.147822612864562,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 27903
    },
    {
      "epoch": 11.148222133439873,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 27904
    },
    {
      "epoch": 11.148621654015182,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 27905
    },
    {
      "epoch": 11.14902117459049,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 27906
    },
    {
      "epoch": 11.149420695165801,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 27907
    },
    {
      "epoch": 11.14982021574111,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 27908
    },
    {
      "epoch": 11.150219736316421,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 27909
    },
    {
      "epoch": 11.15061925689173,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 27910
    },
    {
      "epoch": 11.151018777467039,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 27911
    },
    {
      "epoch": 11.15141829804235,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 27912
    },
    {
      "epoch": 11.151817818617658,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 27913
    },
    {
      "epoch": 11.152217339192969,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 27914
    },
    {
      "epoch": 11.152616859768278,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 27915
    },
    {
      "epoch": 11.153016380343587,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 27916
    },
    {
      "epoch": 11.153415900918898,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 27917
    },
    {
      "epoch": 11.153815421494206,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 27918
    },
    {
      "epoch": 11.154214942069517,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 27919
    },
    {
      "epoch": 11.154614462644826,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 27920
    },
    {
      "epoch": 11.155013983220137,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 27921
    },
    {
      "epoch": 11.155413503795446,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 27922
    },
    {
      "epoch": 11.155813024370755,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 27923
    },
    {
      "epoch": 11.156212544946065,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 27924
    },
    {
      "epoch": 11.156612065521374,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 27925
    },
    {
      "epoch": 11.157011586096685,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 27926
    },
    {
      "epoch": 11.157411106671994,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 27927
    },
    {
      "epoch": 11.157810627247303,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 27928
    },
    {
      "epoch": 11.158210147822613,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 27929
    },
    {
      "epoch": 11.158609668397922,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 27930
    },
    {
      "epoch": 11.159009188973233,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 27931
    },
    {
      "epoch": 11.159408709548542,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 27932
    },
    {
      "epoch": 11.15980823012385,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 27933
    },
    {
      "epoch": 11.160207750699161,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 27934
    },
    {
      "epoch": 11.16060727127447,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 27935
    },
    {
      "epoch": 11.16100679184978,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 27936
    },
    {
      "epoch": 11.16140631242509,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 27937
    },
    {
      "epoch": 11.1618058330004,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 27938
    },
    {
      "epoch": 11.16220535357571,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 27939
    },
    {
      "epoch": 11.162604874151018,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 27940
    },
    {
      "epoch": 11.163004394726329,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 27941
    },
    {
      "epoch": 11.163403915301638,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 27942
    },
    {
      "epoch": 11.163803435876948,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 27943
    },
    {
      "epoch": 11.164202956452257,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 27944
    },
    {
      "epoch": 11.164602477027566,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 27945
    },
    {
      "epoch": 11.165001997602877,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 27946
    },
    {
      "epoch": 11.165401518178186,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 27947
    },
    {
      "epoch": 11.165801038753496,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 27948
    },
    {
      "epoch": 11.166200559328805,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 27949
    },
    {
      "epoch": 11.166600079904114,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 27950
    },
    {
      "epoch": 11.166999600479425,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 27951
    },
    {
      "epoch": 11.167399121054734,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 27952
    },
    {
      "epoch": 11.167798641630045,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 27953
    },
    {
      "epoch": 11.168198162205353,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 27954
    },
    {
      "epoch": 11.168597682780662,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 27955
    },
    {
      "epoch": 11.168997203355973,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 27956
    },
    {
      "epoch": 11.169396723931282,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 27957
    },
    {
      "epoch": 11.169796244506593,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 27958
    },
    {
      "epoch": 11.170195765081901,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 27959
    },
    {
      "epoch": 11.170595285657212,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 27960
    },
    {
      "epoch": 11.170994806232521,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 27961
    },
    {
      "epoch": 11.17139432680783,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 27962
    },
    {
      "epoch": 11.17179384738314,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 27963
    },
    {
      "epoch": 11.17219336795845,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 27964
    },
    {
      "epoch": 11.17259288853376,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 27965
    },
    {
      "epoch": 11.172992409109069,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 27966
    },
    {
      "epoch": 11.173391929684378,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 27967
    },
    {
      "epoch": 11.173791450259689,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 27968
    },
    {
      "epoch": 11.174190970834998,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 27969
    },
    {
      "epoch": 11.174590491410308,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 27970
    },
    {
      "epoch": 11.174990011985617,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 27971
    },
    {
      "epoch": 11.175389532560926,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 27972
    },
    {
      "epoch": 11.175789053136237,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 27973
    },
    {
      "epoch": 11.176188573711546,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 27974
    },
    {
      "epoch": 11.176588094286856,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 27975
    },
    {
      "epoch": 11.176987614862165,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 27976
    },
    {
      "epoch": 11.177387135437476,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 27977
    },
    {
      "epoch": 11.177786656012785,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0513,
      "step": 27978
    },
    {
      "epoch": 11.178186176588094,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 27979
    },
    {
      "epoch": 11.178585697163404,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 27980
    },
    {
      "epoch": 11.178985217738713,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 27981
    },
    {
      "epoch": 11.179384738314024,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 27982
    },
    {
      "epoch": 11.179784258889333,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 27983
    },
    {
      "epoch": 11.180183779464642,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 27984
    },
    {
      "epoch": 11.180583300039952,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 27985
    },
    {
      "epoch": 11.180982820615261,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 27986
    },
    {
      "epoch": 11.181382341190572,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 27987
    },
    {
      "epoch": 11.18178186176588,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 27988
    },
    {
      "epoch": 11.18218138234119,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 27989
    },
    {
      "epoch": 11.1825809029165,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 27990
    },
    {
      "epoch": 11.18298042349181,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 27991
    },
    {
      "epoch": 11.18337994406712,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 27992
    },
    {
      "epoch": 11.183779464642429,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 27993
    },
    {
      "epoch": 11.18417898521774,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 27994
    },
    {
      "epoch": 11.184578505793048,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 27995
    },
    {
      "epoch": 11.184978026368357,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 27996
    },
    {
      "epoch": 11.185377546943668,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 27997
    },
    {
      "epoch": 11.185777067518977,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 27998
    },
    {
      "epoch": 11.186176588094288,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 27999
    },
    {
      "epoch": 11.186576108669597,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 28000
    },
    {
      "epoch": 11.186975629244905,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 28001
    },
    {
      "epoch": 11.187375149820216,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 28002
    },
    {
      "epoch": 11.187774670395525,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 28003
    },
    {
      "epoch": 11.188174190970836,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 28004
    },
    {
      "epoch": 11.188573711546145,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 28005
    },
    {
      "epoch": 11.188973232121453,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 28006
    },
    {
      "epoch": 11.189372752696764,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 28007
    },
    {
      "epoch": 11.189772273272073,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 28008
    },
    {
      "epoch": 11.190171793847384,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 28009
    },
    {
      "epoch": 11.190571314422693,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 28010
    },
    {
      "epoch": 11.190970834998003,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 28011
    },
    {
      "epoch": 11.191370355573312,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 28012
    },
    {
      "epoch": 11.191769876148621,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 28013
    },
    {
      "epoch": 11.192169396723932,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 28014
    },
    {
      "epoch": 11.19256891729924,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 28015
    },
    {
      "epoch": 11.192968437874551,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 28016
    },
    {
      "epoch": 11.19336795844986,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 28017
    },
    {
      "epoch": 11.193767479025169,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 28018
    },
    {
      "epoch": 11.19416699960048,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 28019
    },
    {
      "epoch": 11.194566520175789,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 28020
    },
    {
      "epoch": 11.1949660407511,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 28021
    },
    {
      "epoch": 11.195365561326408,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 28022
    },
    {
      "epoch": 11.195765081901717,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 28023
    },
    {
      "epoch": 11.196164602477028,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 28024
    },
    {
      "epoch": 11.196564123052337,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 28025
    },
    {
      "epoch": 11.196963643627647,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 28026
    },
    {
      "epoch": 11.197363164202956,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 28027
    },
    {
      "epoch": 11.197762684778265,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 28028
    },
    {
      "epoch": 11.198162205353576,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 28029
    },
    {
      "epoch": 11.198561725928885,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 28030
    },
    {
      "epoch": 11.198961246504195,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 28031
    },
    {
      "epoch": 11.199360767079504,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 28032
    },
    {
      "epoch": 11.199760287654815,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 28033
    },
    {
      "epoch": 11.200159808230124,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 28034
    },
    {
      "epoch": 11.200559328805433,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 28035
    },
    {
      "epoch": 11.200958849380743,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 28036
    },
    {
      "epoch": 11.201358369956052,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 28037
    },
    {
      "epoch": 11.201757890531363,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 28038
    },
    {
      "epoch": 11.202157411106672,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 28039
    },
    {
      "epoch": 11.20255693168198,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 28040
    },
    {
      "epoch": 11.202956452257292,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 28041
    },
    {
      "epoch": 11.2033559728326,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 28042
    },
    {
      "epoch": 11.203755493407911,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 28043
    },
    {
      "epoch": 11.20415501398322,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 28044
    },
    {
      "epoch": 11.204554534558529,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 28045
    },
    {
      "epoch": 11.20495405513384,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 28046
    },
    {
      "epoch": 11.205353575709148,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 28047
    },
    {
      "epoch": 11.20575309628446,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 28048
    },
    {
      "epoch": 11.206152616859768,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 28049
    },
    {
      "epoch": 11.206552137435079,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 28050
    },
    {
      "epoch": 11.206951658010388,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 28051
    },
    {
      "epoch": 11.207351178585697,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 28052
    },
    {
      "epoch": 11.207750699161007,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 28053
    },
    {
      "epoch": 11.208150219736316,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 28054
    },
    {
      "epoch": 11.208549740311627,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 28055
    },
    {
      "epoch": 11.208949260886936,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 28056
    },
    {
      "epoch": 11.209348781462245,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 28057
    },
    {
      "epoch": 11.209748302037555,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 28058
    },
    {
      "epoch": 11.210147822612864,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 28059
    },
    {
      "epoch": 11.210547343188175,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 28060
    },
    {
      "epoch": 11.210946863763484,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 28061
    },
    {
      "epoch": 11.211346384338793,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 28062
    },
    {
      "epoch": 11.211745904914103,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 28063
    },
    {
      "epoch": 11.212145425489412,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 28064
    },
    {
      "epoch": 11.212544946064723,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 28065
    },
    {
      "epoch": 11.212944466640032,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 28066
    },
    {
      "epoch": 11.213343987215342,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 28067
    },
    {
      "epoch": 11.213743507790651,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 28068
    },
    {
      "epoch": 11.21414302836596,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 28069
    },
    {
      "epoch": 11.214542548941271,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 28070
    },
    {
      "epoch": 11.21494206951658,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 28071
    },
    {
      "epoch": 11.21534159009189,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 28072
    },
    {
      "epoch": 11.2157411106672,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 28073
    },
    {
      "epoch": 11.216140631242508,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 28074
    },
    {
      "epoch": 11.216540151817819,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 28075
    },
    {
      "epoch": 11.216939672393128,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 28076
    },
    {
      "epoch": 11.217339192968439,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 28077
    },
    {
      "epoch": 11.217738713543747,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 28078
    },
    {
      "epoch": 11.218138234119056,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 28079
    },
    {
      "epoch": 11.218537754694367,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 28080
    },
    {
      "epoch": 11.218937275269676,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 28081
    },
    {
      "epoch": 11.219336795844987,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 28082
    },
    {
      "epoch": 11.219736316420295,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 28083
    },
    {
      "epoch": 11.220135836995606,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 28084
    },
    {
      "epoch": 11.220535357570915,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 28085
    },
    {
      "epoch": 11.220934878146224,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 28086
    },
    {
      "epoch": 11.221334398721535,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 28087
    },
    {
      "epoch": 11.221733919296843,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 28088
    },
    {
      "epoch": 11.222133439872154,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 28089
    },
    {
      "epoch": 11.222532960447463,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 28090
    },
    {
      "epoch": 11.222932481022772,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 28091
    },
    {
      "epoch": 11.223332001598083,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 28092
    },
    {
      "epoch": 11.223731522173392,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 28093
    },
    {
      "epoch": 11.224131042748702,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 28094
    },
    {
      "epoch": 11.224530563324011,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 28095
    },
    {
      "epoch": 11.22493008389932,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 28096
    },
    {
      "epoch": 11.22532960447463,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 28097
    },
    {
      "epoch": 11.22572912504994,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 28098
    },
    {
      "epoch": 11.22612864562525,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 28099
    },
    {
      "epoch": 11.22652816620056,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 28100
    },
    {
      "epoch": 11.22692768677587,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 28101
    },
    {
      "epoch": 11.227327207351179,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 28102
    },
    {
      "epoch": 11.227726727926488,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 28103
    },
    {
      "epoch": 11.228126248501798,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 28104
    },
    {
      "epoch": 11.228525769077107,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 28105
    },
    {
      "epoch": 11.228925289652418,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 28106
    },
    {
      "epoch": 11.229324810227727,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 28107
    },
    {
      "epoch": 11.229724330803036,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 28108
    },
    {
      "epoch": 11.230123851378346,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 28109
    },
    {
      "epoch": 11.230523371953655,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 28110
    },
    {
      "epoch": 11.230922892528966,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 28111
    },
    {
      "epoch": 11.231322413104275,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 28112
    },
    {
      "epoch": 11.231721933679584,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 28113
    },
    {
      "epoch": 11.232121454254894,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 28114
    },
    {
      "epoch": 11.232520974830203,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 28115
    },
    {
      "epoch": 11.232920495405514,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 28116
    },
    {
      "epoch": 11.233320015980823,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 28117
    },
    {
      "epoch": 11.233719536556134,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 28118
    },
    {
      "epoch": 11.234119057131442,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 28119
    },
    {
      "epoch": 11.234518577706751,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 28120
    },
    {
      "epoch": 11.234918098282062,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 28121
    },
    {
      "epoch": 11.235317618857371,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 28122
    },
    {
      "epoch": 11.235717139432682,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 28123
    },
    {
      "epoch": 11.23611666000799,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 28124
    },
    {
      "epoch": 11.2365161805833,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 28125
    },
    {
      "epoch": 11.23691570115861,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 28126
    },
    {
      "epoch": 11.237315221733919,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 28127
    },
    {
      "epoch": 11.23771474230923,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 28128
    },
    {
      "epoch": 11.238114262884539,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 28129
    },
    {
      "epoch": 11.238513783459847,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 28130
    },
    {
      "epoch": 11.238913304035158,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 28131
    },
    {
      "epoch": 11.239312824610467,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 28132
    },
    {
      "epoch": 11.239712345185778,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 28133
    },
    {
      "epoch": 11.240111865761087,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 28134
    },
    {
      "epoch": 11.240511386336395,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 28135
    },
    {
      "epoch": 11.240910906911706,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 28136
    },
    {
      "epoch": 11.241310427487015,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 28137
    },
    {
      "epoch": 11.241709948062326,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 28138
    },
    {
      "epoch": 11.242109468637635,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 28139
    },
    {
      "epoch": 11.242508989212945,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 28140
    },
    {
      "epoch": 11.242908509788254,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 28141
    },
    {
      "epoch": 11.243308030363563,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 28142
    },
    {
      "epoch": 11.243707550938874,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 28143
    },
    {
      "epoch": 11.244107071514183,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 28144
    },
    {
      "epoch": 11.244506592089493,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 28145
    },
    {
      "epoch": 11.244906112664802,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 28146
    },
    {
      "epoch": 11.245305633240111,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 28147
    },
    {
      "epoch": 11.245705153815422,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 28148
    },
    {
      "epoch": 11.24610467439073,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 28149
    },
    {
      "epoch": 11.246504194966041,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 28150
    },
    {
      "epoch": 11.24690371554135,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 28151
    },
    {
      "epoch": 11.24730323611666,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 28152
    },
    {
      "epoch": 11.24770275669197,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 28153
    },
    {
      "epoch": 11.248102277267279,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 28154
    },
    {
      "epoch": 11.24850179784259,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 28155
    },
    {
      "epoch": 11.248901318417898,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 28156
    },
    {
      "epoch": 11.249300838993209,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 28157
    },
    {
      "epoch": 11.249700359568518,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 28158
    },
    {
      "epoch": 11.250099880143827,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 28159
    },
    {
      "epoch": 11.250499400719137,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 28160
    },
    {
      "epoch": 11.250898921294446,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 28161
    },
    {
      "epoch": 11.251298441869757,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 28162
    },
    {
      "epoch": 11.251697962445066,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 28163
    },
    {
      "epoch": 11.252097483020375,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 28164
    },
    {
      "epoch": 11.252497003595685,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 28165
    },
    {
      "epoch": 11.252896524170994,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 28166
    },
    {
      "epoch": 11.253296044746305,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 28167
    },
    {
      "epoch": 11.253695565321614,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 28168
    },
    {
      "epoch": 11.254095085896923,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 28169
    },
    {
      "epoch": 11.254494606472234,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 28170
    },
    {
      "epoch": 11.254894127047542,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 28171
    },
    {
      "epoch": 11.255293647622853,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 28172
    },
    {
      "epoch": 11.255693168198162,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0105,
      "step": 28173
    },
    {
      "epoch": 11.256092688773473,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 28174
    },
    {
      "epoch": 11.256492209348782,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 28175
    },
    {
      "epoch": 11.25689172992409,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 28176
    },
    {
      "epoch": 11.257291250499401,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 28177
    },
    {
      "epoch": 11.25769077107471,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 28178
    },
    {
      "epoch": 11.25809029165002,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 28179
    },
    {
      "epoch": 11.25848981222533,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 28180
    },
    {
      "epoch": 11.258889332800639,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 28181
    },
    {
      "epoch": 11.25928885337595,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 28182
    },
    {
      "epoch": 11.259688373951258,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 28183
    },
    {
      "epoch": 11.260087894526569,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 28184
    },
    {
      "epoch": 11.260487415101878,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 28185
    },
    {
      "epoch": 11.260886935677187,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 28186
    },
    {
      "epoch": 11.261286456252497,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 28187
    },
    {
      "epoch": 11.261685976827806,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 28188
    },
    {
      "epoch": 11.262085497403117,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 28189
    },
    {
      "epoch": 11.262485017978426,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 28190
    },
    {
      "epoch": 11.262884538553735,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 28191
    },
    {
      "epoch": 11.263284059129045,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 28192
    },
    {
      "epoch": 11.263683579704354,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 28193
    },
    {
      "epoch": 11.264083100279665,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 28194
    },
    {
      "epoch": 11.264482620854974,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 28195
    },
    {
      "epoch": 11.264882141430284,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 28196
    },
    {
      "epoch": 11.265281662005593,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 28197
    },
    {
      "epoch": 11.265681182580902,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 28198
    },
    {
      "epoch": 11.266080703156213,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 28199
    },
    {
      "epoch": 11.266480223731522,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 28200
    },
    {
      "epoch": 11.266879744306832,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 28201
    },
    {
      "epoch": 11.267279264882141,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 28202
    },
    {
      "epoch": 11.26767878545745,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 28203
    },
    {
      "epoch": 11.268078306032761,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 28204
    },
    {
      "epoch": 11.26847782660807,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 28205
    },
    {
      "epoch": 11.26887734718338,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 28206
    },
    {
      "epoch": 11.26927686775869,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 28207
    },
    {
      "epoch": 11.269676388333998,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 28208
    },
    {
      "epoch": 11.270075908909309,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 28209
    },
    {
      "epoch": 11.270475429484618,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 28210
    },
    {
      "epoch": 11.270874950059929,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 28211
    },
    {
      "epoch": 11.271274470635237,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 28212
    },
    {
      "epoch": 11.271673991210548,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 28213
    },
    {
      "epoch": 11.272073511785857,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 28214
    },
    {
      "epoch": 11.272473032361166,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 28215
    },
    {
      "epoch": 11.272872552936477,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 28216
    },
    {
      "epoch": 11.273272073511786,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 28217
    },
    {
      "epoch": 11.273671594087096,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 28218
    },
    {
      "epoch": 11.274071114662405,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 28219
    },
    {
      "epoch": 11.274470635237714,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 28220
    },
    {
      "epoch": 11.274870155813025,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 28221
    },
    {
      "epoch": 11.275269676388334,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 28222
    },
    {
      "epoch": 11.275669196963644,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 28223
    },
    {
      "epoch": 11.276068717538953,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 28224
    },
    {
      "epoch": 11.276468238114262,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 28225
    },
    {
      "epoch": 11.276867758689573,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 28226
    },
    {
      "epoch": 11.277267279264882,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 28227
    },
    {
      "epoch": 11.277666799840192,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 28228
    },
    {
      "epoch": 11.278066320415501,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 28229
    },
    {
      "epoch": 11.278465840990812,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 28230
    },
    {
      "epoch": 11.27886536156612,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 28231
    },
    {
      "epoch": 11.27926488214143,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 28232
    },
    {
      "epoch": 11.27966440271674,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 28233
    },
    {
      "epoch": 11.28006392329205,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 28234
    },
    {
      "epoch": 11.28046344386736,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 28235
    },
    {
      "epoch": 11.280862964442669,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 28236
    },
    {
      "epoch": 11.281262485017978,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 28237
    },
    {
      "epoch": 11.281662005593288,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 28238
    },
    {
      "epoch": 11.282061526168597,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 28239
    },
    {
      "epoch": 11.282461046743908,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 28240
    },
    {
      "epoch": 11.282860567319217,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 28241
    },
    {
      "epoch": 11.283260087894526,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 28242
    },
    {
      "epoch": 11.283659608469836,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 28243
    },
    {
      "epoch": 11.284059129045145,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 28244
    },
    {
      "epoch": 11.284458649620456,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 28245
    },
    {
      "epoch": 11.284858170195765,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 28246
    },
    {
      "epoch": 11.285257690771076,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 28247
    },
    {
      "epoch": 11.285657211346384,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 28248
    },
    {
      "epoch": 11.286056731921693,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 28249
    },
    {
      "epoch": 11.286456252497004,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 28250
    },
    {
      "epoch": 11.286855773072313,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 28251
    },
    {
      "epoch": 11.287255293647624,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 28252
    },
    {
      "epoch": 11.287654814222932,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 28253
    },
    {
      "epoch": 11.288054334798241,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 28254
    },
    {
      "epoch": 11.288453855373552,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 28255
    },
    {
      "epoch": 11.288853375948861,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 28256
    },
    {
      "epoch": 11.289252896524172,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 28257
    },
    {
      "epoch": 11.28965241709948,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 28258
    },
    {
      "epoch": 11.29005193767479,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 28259
    },
    {
      "epoch": 11.2904514582501,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 28260
    },
    {
      "epoch": 11.290850978825409,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 28261
    },
    {
      "epoch": 11.29125049940072,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 28262
    },
    {
      "epoch": 11.291650019976029,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 28263
    },
    {
      "epoch": 11.29204954055134,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 28264
    },
    {
      "epoch": 11.292449061126648,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 28265
    },
    {
      "epoch": 11.292848581701957,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 28266
    },
    {
      "epoch": 11.293248102277268,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 28267
    },
    {
      "epoch": 11.293647622852577,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 28268
    },
    {
      "epoch": 11.294047143427887,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0514,
      "step": 28269
    },
    {
      "epoch": 11.294446664003196,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 28270
    },
    {
      "epoch": 11.294846184578505,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 28271
    },
    {
      "epoch": 11.295245705153816,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 28272
    },
    {
      "epoch": 11.295645225729125,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 28273
    },
    {
      "epoch": 11.296044746304435,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 28274
    },
    {
      "epoch": 11.296444266879744,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 28275
    },
    {
      "epoch": 11.296843787455053,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 28276
    },
    {
      "epoch": 11.297243308030364,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0121,
      "step": 28277
    },
    {
      "epoch": 11.297642828605673,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 28278
    },
    {
      "epoch": 11.298042349180983,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 28279
    },
    {
      "epoch": 11.298441869756292,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 28280
    },
    {
      "epoch": 11.298841390331603,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 28281
    },
    {
      "epoch": 11.299240910906912,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0494,
      "step": 28282
    },
    {
      "epoch": 11.29964043148222,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 28283
    },
    {
      "epoch": 11.300039952057531,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 28284
    },
    {
      "epoch": 11.30043947263284,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 28285
    },
    {
      "epoch": 11.300838993208151,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 28286
    },
    {
      "epoch": 11.30123851378346,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 28287
    },
    {
      "epoch": 11.301638034358769,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 28288
    },
    {
      "epoch": 11.30203755493408,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 28289
    },
    {
      "epoch": 11.302437075509388,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 28290
    },
    {
      "epoch": 11.302836596084699,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 28291
    },
    {
      "epoch": 11.303236116660008,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 28292
    },
    {
      "epoch": 11.303635637235317,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 28293
    },
    {
      "epoch": 11.304035157810628,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 28294
    },
    {
      "epoch": 11.304434678385936,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 28295
    },
    {
      "epoch": 11.304834198961247,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 28296
    },
    {
      "epoch": 11.305233719536556,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 28297
    },
    {
      "epoch": 11.305633240111867,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 28298
    },
    {
      "epoch": 11.306032760687176,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 28299
    },
    {
      "epoch": 11.306432281262484,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 28300
    },
    {
      "epoch": 11.306831801837795,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 28301
    },
    {
      "epoch": 11.307231322413104,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 28302
    },
    {
      "epoch": 11.307630842988415,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 28303
    },
    {
      "epoch": 11.308030363563724,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 28304
    },
    {
      "epoch": 11.308429884139032,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 28305
    },
    {
      "epoch": 11.308829404714343,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 28306
    },
    {
      "epoch": 11.309228925289652,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 28307
    },
    {
      "epoch": 11.309628445864963,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 28308
    },
    {
      "epoch": 11.310027966440272,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 28309
    },
    {
      "epoch": 11.31042748701558,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 28310
    },
    {
      "epoch": 11.310827007590891,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 28311
    },
    {
      "epoch": 11.3112265281662,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 28312
    },
    {
      "epoch": 11.31162604874151,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 28313
    },
    {
      "epoch": 11.31202556931682,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 28314
    },
    {
      "epoch": 11.31242508989213,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 28315
    },
    {
      "epoch": 11.31282461046744,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 28316
    },
    {
      "epoch": 11.313224131042748,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 28317
    },
    {
      "epoch": 11.313623651618059,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 28318
    },
    {
      "epoch": 11.314023172193368,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 28319
    },
    {
      "epoch": 11.314422692768678,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 28320
    },
    {
      "epoch": 11.314822213343987,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 28321
    },
    {
      "epoch": 11.315221733919296,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 28322
    },
    {
      "epoch": 11.315621254494607,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 28323
    },
    {
      "epoch": 11.316020775069916,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 28324
    },
    {
      "epoch": 11.316420295645226,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 28325
    },
    {
      "epoch": 11.316819816220535,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 28326
    },
    {
      "epoch": 11.317219336795844,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 28327
    },
    {
      "epoch": 11.317618857371155,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 28328
    },
    {
      "epoch": 11.318018377946464,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 28329
    },
    {
      "epoch": 11.318417898521774,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 28330
    },
    {
      "epoch": 11.318817419097083,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 28331
    },
    {
      "epoch": 11.319216939672392,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 28332
    },
    {
      "epoch": 11.319616460247703,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 28333
    },
    {
      "epoch": 11.320015980823012,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 28334
    },
    {
      "epoch": 11.320415501398323,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 28335
    },
    {
      "epoch": 11.320815021973631,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 28336
    },
    {
      "epoch": 11.321214542548942,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 28337
    },
    {
      "epoch": 11.321614063124251,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 28338
    },
    {
      "epoch": 11.32201358369956,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 28339
    },
    {
      "epoch": 11.32241310427487,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 28340
    },
    {
      "epoch": 11.32281262485018,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 28341
    },
    {
      "epoch": 11.32321214542549,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 28342
    },
    {
      "epoch": 11.323611666000799,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 28343
    },
    {
      "epoch": 11.324011186576108,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 28344
    },
    {
      "epoch": 11.324410707151419,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 28345
    },
    {
      "epoch": 11.324810227726728,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 28346
    },
    {
      "epoch": 11.325209748302038,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 28347
    },
    {
      "epoch": 11.325609268877347,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 28348
    },
    {
      "epoch": 11.326008789452656,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 28349
    },
    {
      "epoch": 11.326408310027967,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 28350
    },
    {
      "epoch": 11.326807830603276,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 28351
    },
    {
      "epoch": 11.327207351178586,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 28352
    },
    {
      "epoch": 11.327606871753895,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 28353
    },
    {
      "epoch": 11.328006392329206,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 28354
    },
    {
      "epoch": 11.328405912904515,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 28355
    },
    {
      "epoch": 11.328805433479824,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 28356
    },
    {
      "epoch": 11.329204954055134,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 28357
    },
    {
      "epoch": 11.329604474630443,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 28358
    },
    {
      "epoch": 11.330003995205754,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 28359
    },
    {
      "epoch": 11.330403515781063,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 28360
    },
    {
      "epoch": 11.330803036356372,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 28361
    },
    {
      "epoch": 11.331202556931682,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 28362
    },
    {
      "epoch": 11.331602077506991,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 28363
    },
    {
      "epoch": 11.332001598082302,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 28364
    },
    {
      "epoch": 11.33240111865761,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 28365
    },
    {
      "epoch": 11.33280063923292,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 28366
    },
    {
      "epoch": 11.33320015980823,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 28367
    },
    {
      "epoch": 11.33359968038354,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 28368
    },
    {
      "epoch": 11.33399920095885,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 28369
    },
    {
      "epoch": 11.334398721534159,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 28370
    },
    {
      "epoch": 11.334798242109468,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 28371
    },
    {
      "epoch": 11.335197762684778,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 28372
    },
    {
      "epoch": 11.335597283260087,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 28373
    },
    {
      "epoch": 11.335996803835398,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 28374
    },
    {
      "epoch": 11.336396324410707,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 28375
    },
    {
      "epoch": 11.336795844986018,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 28376
    },
    {
      "epoch": 11.337195365561326,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 28377
    },
    {
      "epoch": 11.337594886136635,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 28378
    },
    {
      "epoch": 11.337994406711946,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 28379
    },
    {
      "epoch": 11.338393927287255,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 28380
    },
    {
      "epoch": 11.338793447862566,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 28381
    },
    {
      "epoch": 11.339192968437874,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 28382
    },
    {
      "epoch": 11.339592489013183,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 28383
    },
    {
      "epoch": 11.339992009588494,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 28384
    },
    {
      "epoch": 11.340391530163803,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 28385
    },
    {
      "epoch": 11.340791050739114,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 28386
    },
    {
      "epoch": 11.341190571314423,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 28387
    },
    {
      "epoch": 11.341590091889731,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 28388
    },
    {
      "epoch": 11.341989612465042,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 28389
    },
    {
      "epoch": 11.342389133040351,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 28390
    },
    {
      "epoch": 11.342788653615662,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 28391
    },
    {
      "epoch": 11.34318817419097,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 28392
    },
    {
      "epoch": 11.343587694766281,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 28393
    },
    {
      "epoch": 11.34398721534159,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 28394
    },
    {
      "epoch": 11.344386735916899,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 28395
    },
    {
      "epoch": 11.34478625649221,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 28396
    },
    {
      "epoch": 11.345185777067519,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 28397
    },
    {
      "epoch": 11.34558529764283,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 28398
    },
    {
      "epoch": 11.345984818218138,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 28399
    },
    {
      "epoch": 11.346384338793447,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 28400
    },
    {
      "epoch": 11.346783859368758,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 28401
    },
    {
      "epoch": 11.347183379944067,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 28402
    },
    {
      "epoch": 11.347582900519377,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 28403
    },
    {
      "epoch": 11.347982421094686,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 28404
    },
    {
      "epoch": 11.348381941669995,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 28405
    },
    {
      "epoch": 11.348781462245306,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 28406
    },
    {
      "epoch": 11.349180982820615,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 28407
    },
    {
      "epoch": 11.349580503395925,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 28408
    },
    {
      "epoch": 11.349980023971234,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 28409
    },
    {
      "epoch": 11.350379544546545,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 28410
    },
    {
      "epoch": 11.350779065121854,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 28411
    },
    {
      "epoch": 11.351178585697163,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 28412
    },
    {
      "epoch": 11.351578106272473,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 28413
    },
    {
      "epoch": 11.351977626847782,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 28414
    },
    {
      "epoch": 11.352377147423093,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 28415
    },
    {
      "epoch": 11.352776667998402,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 28416
    },
    {
      "epoch": 11.35317618857371,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 28417
    },
    {
      "epoch": 11.353575709149021,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 28418
    },
    {
      "epoch": 11.35397522972433,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 28419
    },
    {
      "epoch": 11.354374750299641,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 28420
    },
    {
      "epoch": 11.35477427087495,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 28421
    },
    {
      "epoch": 11.355173791450259,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 28422
    },
    {
      "epoch": 11.35557331202557,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 28423
    },
    {
      "epoch": 11.355972832600878,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 28424
    },
    {
      "epoch": 11.356372353176189,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 28425
    },
    {
      "epoch": 11.356771873751498,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 28426
    },
    {
      "epoch": 11.357171394326809,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 28427
    },
    {
      "epoch": 11.357570914902118,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 28428
    },
    {
      "epoch": 11.357970435477426,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 28429
    },
    {
      "epoch": 11.358369956052737,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 28430
    },
    {
      "epoch": 11.358769476628046,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 28431
    },
    {
      "epoch": 11.359168997203357,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 28432
    },
    {
      "epoch": 11.359568517778666,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 28433
    },
    {
      "epoch": 11.359968038353975,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 28434
    },
    {
      "epoch": 11.360367558929285,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 28435
    },
    {
      "epoch": 11.360767079504594,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 28436
    },
    {
      "epoch": 11.361166600079905,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 28437
    },
    {
      "epoch": 11.361566120655214,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 28438
    },
    {
      "epoch": 11.361965641230523,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 28439
    },
    {
      "epoch": 11.362365161805833,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 28440
    },
    {
      "epoch": 11.362764682381142,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 28441
    },
    {
      "epoch": 11.363164202956453,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 28442
    },
    {
      "epoch": 11.363563723531762,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 28443
    },
    {
      "epoch": 11.363963244107072,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 28444
    },
    {
      "epoch": 11.364362764682381,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 28445
    },
    {
      "epoch": 11.36476228525769,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 28446
    },
    {
      "epoch": 11.365161805833,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 28447
    },
    {
      "epoch": 11.36556132640831,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 28448
    },
    {
      "epoch": 11.36596084698362,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 28449
    },
    {
      "epoch": 11.36636036755893,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 28450
    },
    {
      "epoch": 11.366759888134238,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 28451
    },
    {
      "epoch": 11.367159408709549,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 28452
    },
    {
      "epoch": 11.367558929284858,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 28453
    },
    {
      "epoch": 11.367958449860168,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 28454
    },
    {
      "epoch": 11.368357970435477,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 28455
    },
    {
      "epoch": 11.368757491010786,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 28456
    },
    {
      "epoch": 11.369157011586097,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 28457
    },
    {
      "epoch": 11.369556532161406,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 28458
    },
    {
      "epoch": 11.369956052736717,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 28459
    },
    {
      "epoch": 11.370355573312025,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 28460
    },
    {
      "epoch": 11.370755093887336,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 28461
    },
    {
      "epoch": 11.371154614462645,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 28462
    },
    {
      "epoch": 11.371554135037954,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 28463
    },
    {
      "epoch": 11.371953655613265,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 28464
    },
    {
      "epoch": 11.372353176188573,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 28465
    },
    {
      "epoch": 11.372752696763884,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 28466
    },
    {
      "epoch": 11.373152217339193,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 28467
    },
    {
      "epoch": 11.373551737914502,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 28468
    },
    {
      "epoch": 11.373951258489813,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 28469
    },
    {
      "epoch": 11.374350779065121,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 28470
    },
    {
      "epoch": 11.374750299640432,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 28471
    },
    {
      "epoch": 11.375149820215741,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 28472
    },
    {
      "epoch": 11.37554934079105,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 28473
    },
    {
      "epoch": 11.37594886136636,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 28474
    },
    {
      "epoch": 11.37634838194167,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 28475
    },
    {
      "epoch": 11.37674790251698,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 28476
    },
    {
      "epoch": 11.377147423092289,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 28477
    },
    {
      "epoch": 11.3775469436676,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 28478
    },
    {
      "epoch": 11.377946464242909,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 28479
    },
    {
      "epoch": 11.378345984818218,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 28480
    },
    {
      "epoch": 11.378745505393528,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 28481
    },
    {
      "epoch": 11.379145025968837,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 28482
    },
    {
      "epoch": 11.379544546544148,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 28483
    },
    {
      "epoch": 11.379944067119457,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 28484
    },
    {
      "epoch": 11.380343587694766,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 28485
    },
    {
      "epoch": 11.380743108270076,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 28486
    },
    {
      "epoch": 11.381142628845385,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 28487
    },
    {
      "epoch": 11.381542149420696,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 28488
    },
    {
      "epoch": 11.381941669996005,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 28489
    },
    {
      "epoch": 11.382341190571314,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 28490
    },
    {
      "epoch": 11.382740711146624,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 28491
    },
    {
      "epoch": 11.383140231721933,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 28492
    },
    {
      "epoch": 11.383539752297244,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 28493
    },
    {
      "epoch": 11.383939272872553,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 28494
    },
    {
      "epoch": 11.384338793447863,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 28495
    },
    {
      "epoch": 11.384738314023172,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 28496
    },
    {
      "epoch": 11.385137834598481,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 28497
    },
    {
      "epoch": 11.385537355173792,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 28498
    },
    {
      "epoch": 11.3859368757491,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 28499
    },
    {
      "epoch": 11.386336396324412,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 28500
    },
    {
      "epoch": 11.38673591689972,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 28501
    },
    {
      "epoch": 11.38713543747503,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 28502
    },
    {
      "epoch": 11.38753495805034,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 28503
    },
    {
      "epoch": 11.387934478625649,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 28504
    },
    {
      "epoch": 11.38833399920096,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 28505
    },
    {
      "epoch": 11.388733519776268,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 28506
    },
    {
      "epoch": 11.389133040351577,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 28507
    },
    {
      "epoch": 11.389532560926888,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 28508
    },
    {
      "epoch": 11.389932081502197,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 28509
    },
    {
      "epoch": 11.390331602077508,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 28510
    },
    {
      "epoch": 11.390731122652817,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 28511
    },
    {
      "epoch": 11.391130643228125,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 28512
    },
    {
      "epoch": 11.391530163803436,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 28513
    },
    {
      "epoch": 11.391929684378745,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 28514
    },
    {
      "epoch": 11.392329204954056,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 28515
    },
    {
      "epoch": 11.392728725529365,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 28516
    },
    {
      "epoch": 11.393128246104675,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 28517
    },
    {
      "epoch": 11.393527766679984,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 28518
    },
    {
      "epoch": 11.393927287255293,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 28519
    },
    {
      "epoch": 11.394326807830604,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 28520
    },
    {
      "epoch": 11.394726328405913,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 28521
    },
    {
      "epoch": 11.395125848981223,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 28522
    },
    {
      "epoch": 11.395525369556532,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 28523
    },
    {
      "epoch": 11.395924890131841,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 28524
    },
    {
      "epoch": 11.396324410707152,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 28525
    },
    {
      "epoch": 11.39672393128246,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 28526
    },
    {
      "epoch": 11.397123451857771,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 28527
    },
    {
      "epoch": 11.39752297243308,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 28528
    },
    {
      "epoch": 11.397922493008389,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 28529
    },
    {
      "epoch": 11.3983220135837,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 28530
    },
    {
      "epoch": 11.398721534159009,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 28531
    },
    {
      "epoch": 11.39912105473432,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 28532
    },
    {
      "epoch": 11.399520575309628,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 28533
    },
    {
      "epoch": 11.399920095884939,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 28534
    },
    {
      "epoch": 11.400319616460248,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 28535
    },
    {
      "epoch": 11.400719137035557,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 28536
    },
    {
      "epoch": 11.401118657610867,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 28537
    },
    {
      "epoch": 11.401518178186176,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 28538
    },
    {
      "epoch": 11.401917698761487,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 28539
    },
    {
      "epoch": 11.402317219336796,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 28540
    },
    {
      "epoch": 11.402716739912105,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 28541
    },
    {
      "epoch": 11.403116260487415,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 28542
    },
    {
      "epoch": 11.403515781062724,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 28543
    },
    {
      "epoch": 11.403915301638035,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 28544
    },
    {
      "epoch": 11.404314822213344,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 28545
    },
    {
      "epoch": 11.404714342788653,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 28546
    },
    {
      "epoch": 11.405113863363963,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 28547
    },
    {
      "epoch": 11.405513383939272,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 28548
    },
    {
      "epoch": 11.405912904514583,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 28549
    },
    {
      "epoch": 11.406312425089892,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 28550
    },
    {
      "epoch": 11.4067119456652,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 28551
    },
    {
      "epoch": 11.407111466240512,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 28552
    },
    {
      "epoch": 11.40751098681582,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 28553
    },
    {
      "epoch": 11.407910507391131,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 28554
    },
    {
      "epoch": 11.40831002796644,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 28555
    },
    {
      "epoch": 11.40870954854175,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 28556
    },
    {
      "epoch": 11.40910906911706,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 28557
    },
    {
      "epoch": 11.409508589692368,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 28558
    },
    {
      "epoch": 11.40990811026768,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 28559
    },
    {
      "epoch": 11.410307630842988,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 28560
    },
    {
      "epoch": 11.410707151418299,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 28561
    },
    {
      "epoch": 11.411106671993608,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 28562
    },
    {
      "epoch": 11.411506192568917,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 28563
    },
    {
      "epoch": 11.411905713144227,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 28564
    },
    {
      "epoch": 11.412305233719536,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 28565
    },
    {
      "epoch": 11.412704754294847,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 28566
    },
    {
      "epoch": 11.413104274870156,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 28567
    },
    {
      "epoch": 11.413503795445465,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 28568
    },
    {
      "epoch": 11.413903316020775,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 28569
    },
    {
      "epoch": 11.414302836596084,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 28570
    },
    {
      "epoch": 11.414702357171395,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 28571
    },
    {
      "epoch": 11.415101877746704,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 28572
    },
    {
      "epoch": 11.415501398322014,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 28573
    },
    {
      "epoch": 11.415900918897323,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 28574
    },
    {
      "epoch": 11.416300439472632,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 28575
    },
    {
      "epoch": 11.416699960047943,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 28576
    },
    {
      "epoch": 11.417099480623252,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 28577
    },
    {
      "epoch": 11.417499001198562,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 28578
    },
    {
      "epoch": 11.417898521773871,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 28579
    },
    {
      "epoch": 11.41829804234918,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 28580
    },
    {
      "epoch": 11.418697562924491,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 28581
    },
    {
      "epoch": 11.4190970834998,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 28582
    },
    {
      "epoch": 11.41949660407511,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 28583
    },
    {
      "epoch": 11.41989612465042,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 28584
    },
    {
      "epoch": 11.420295645225728,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 28585
    },
    {
      "epoch": 11.420695165801039,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 28586
    },
    {
      "epoch": 11.421094686376348,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 28587
    },
    {
      "epoch": 11.421494206951659,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 28588
    },
    {
      "epoch": 11.421893727526967,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 28589
    },
    {
      "epoch": 11.422293248102278,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 28590
    },
    {
      "epoch": 11.422692768677587,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 28591
    },
    {
      "epoch": 11.423092289252896,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 28592
    },
    {
      "epoch": 11.423491809828207,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 28593
    },
    {
      "epoch": 11.423891330403515,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 28594
    },
    {
      "epoch": 11.424290850978826,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 28595
    },
    {
      "epoch": 11.424690371554135,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 28596
    },
    {
      "epoch": 11.425089892129444,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 28597
    },
    {
      "epoch": 11.425489412704755,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 28598
    },
    {
      "epoch": 11.425888933280063,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 28599
    },
    {
      "epoch": 11.426288453855374,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 28600
    },
    {
      "epoch": 11.426687974430683,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 28601
    },
    {
      "epoch": 11.427087495005992,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 28602
    },
    {
      "epoch": 11.427487015581303,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 28603
    },
    {
      "epoch": 11.427886536156612,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 28604
    },
    {
      "epoch": 11.428286056731922,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 28605
    },
    {
      "epoch": 11.428685577307231,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 28606
    },
    {
      "epoch": 11.429085097882542,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 28607
    },
    {
      "epoch": 11.42948461845785,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 28608
    },
    {
      "epoch": 11.42988413903316,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 28609
    },
    {
      "epoch": 11.43028365960847,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 28610
    },
    {
      "epoch": 11.43068318018378,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 28611
    },
    {
      "epoch": 11.43108270075909,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 28612
    },
    {
      "epoch": 11.431482221334399,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 28613
    },
    {
      "epoch": 11.431881741909708,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 28614
    },
    {
      "epoch": 11.432281262485018,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 28615
    },
    {
      "epoch": 11.432680783060327,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 28616
    },
    {
      "epoch": 11.433080303635638,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 28617
    },
    {
      "epoch": 11.433479824210947,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 28618
    },
    {
      "epoch": 11.433879344786256,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 28619
    },
    {
      "epoch": 11.434278865361566,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 28620
    },
    {
      "epoch": 11.434678385936875,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 28621
    },
    {
      "epoch": 11.435077906512186,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 28622
    },
    {
      "epoch": 11.435477427087495,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 28623
    },
    {
      "epoch": 11.435876947662805,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 28624
    },
    {
      "epoch": 11.436276468238114,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 28625
    },
    {
      "epoch": 11.436675988813423,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 28626
    },
    {
      "epoch": 11.437075509388734,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 28627
    },
    {
      "epoch": 11.437475029964043,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 28628
    },
    {
      "epoch": 11.437874550539354,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 28629
    },
    {
      "epoch": 11.438274071114662,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 28630
    },
    {
      "epoch": 11.438673591689971,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 28631
    },
    {
      "epoch": 11.439073112265282,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 28632
    },
    {
      "epoch": 11.439472632840591,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 28633
    },
    {
      "epoch": 11.439872153415902,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 28634
    },
    {
      "epoch": 11.44027167399121,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 28635
    },
    {
      "epoch": 11.44067119456652,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 28636
    },
    {
      "epoch": 11.44107071514183,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 28637
    },
    {
      "epoch": 11.441470235717139,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 28638
    },
    {
      "epoch": 11.44186975629245,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 28639
    },
    {
      "epoch": 11.442269276867759,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 28640
    },
    {
      "epoch": 11.44266879744307,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 28641
    },
    {
      "epoch": 11.443068318018378,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 28642
    },
    {
      "epoch": 11.443467838593687,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 28643
    },
    {
      "epoch": 11.443867359168998,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 28644
    },
    {
      "epoch": 11.444266879744307,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 28645
    },
    {
      "epoch": 11.444666400319617,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 28646
    },
    {
      "epoch": 11.445065920894926,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 28647
    },
    {
      "epoch": 11.445465441470235,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 28648
    },
    {
      "epoch": 11.445864962045546,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 28649
    },
    {
      "epoch": 11.446264482620855,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 28650
    },
    {
      "epoch": 11.446664003196165,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 28651
    },
    {
      "epoch": 11.447063523771474,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 28652
    },
    {
      "epoch": 11.447463044346783,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 28653
    },
    {
      "epoch": 11.447862564922094,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 28654
    },
    {
      "epoch": 11.448262085497403,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 28655
    },
    {
      "epoch": 11.448661606072713,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 28656
    },
    {
      "epoch": 11.449061126648022,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 28657
    },
    {
      "epoch": 11.449460647223333,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 28658
    },
    {
      "epoch": 11.449860167798642,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 28659
    },
    {
      "epoch": 11.45025968837395,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 28660
    },
    {
      "epoch": 11.450659208949261,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 28661
    },
    {
      "epoch": 11.45105872952457,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 28662
    },
    {
      "epoch": 11.451458250099881,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 28663
    },
    {
      "epoch": 11.45185777067519,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 28664
    },
    {
      "epoch": 11.452257291250499,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 28665
    },
    {
      "epoch": 11.45265681182581,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 28666
    },
    {
      "epoch": 11.453056332401118,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 28667
    },
    {
      "epoch": 11.453455852976429,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 28668
    },
    {
      "epoch": 11.453855373551738,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 28669
    },
    {
      "epoch": 11.454254894127047,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 28670
    },
    {
      "epoch": 11.454654414702357,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 28671
    },
    {
      "epoch": 11.455053935277666,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 28672
    },
    {
      "epoch": 11.455453455852977,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 28673
    },
    {
      "epoch": 11.455852976428286,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 28674
    },
    {
      "epoch": 11.456252497003595,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 28675
    },
    {
      "epoch": 11.456652017578906,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 28676
    },
    {
      "epoch": 11.457051538154214,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 28677
    },
    {
      "epoch": 11.457451058729525,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 28678
    },
    {
      "epoch": 11.457850579304834,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 28679
    },
    {
      "epoch": 11.458250099880145,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 28680
    },
    {
      "epoch": 11.458649620455454,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 28681
    },
    {
      "epoch": 11.459049141030762,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 28682
    },
    {
      "epoch": 11.459448661606073,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 28683
    },
    {
      "epoch": 11.459848182181382,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 28684
    },
    {
      "epoch": 11.460247702756693,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 28685
    },
    {
      "epoch": 11.460647223332002,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 28686
    },
    {
      "epoch": 11.46104674390731,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 28687
    },
    {
      "epoch": 11.461446264482621,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 28688
    },
    {
      "epoch": 11.46184578505793,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 28689
    },
    {
      "epoch": 11.46224530563324,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 28690
    },
    {
      "epoch": 11.46264482620855,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 28691
    },
    {
      "epoch": 11.463044346783859,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 28692
    },
    {
      "epoch": 11.46344386735917,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 28693
    },
    {
      "epoch": 11.463843387934478,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 28694
    },
    {
      "epoch": 11.464242908509789,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 28695
    },
    {
      "epoch": 11.464642429085098,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 28696
    },
    {
      "epoch": 11.465041949660408,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 28697
    },
    {
      "epoch": 11.465441470235717,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 28698
    },
    {
      "epoch": 11.465840990811026,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 28699
    },
    {
      "epoch": 11.466240511386337,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 28700
    },
    {
      "epoch": 11.466640031961646,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 28701
    },
    {
      "epoch": 11.467039552536956,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 28702
    },
    {
      "epoch": 11.467439073112265,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 28703
    },
    {
      "epoch": 11.467838593687574,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 28704
    },
    {
      "epoch": 11.468238114262885,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 28705
    },
    {
      "epoch": 11.468637634838194,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 28706
    },
    {
      "epoch": 11.469037155413504,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 28707
    },
    {
      "epoch": 11.469436675988813,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 28708
    },
    {
      "epoch": 11.469836196564122,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 28709
    },
    {
      "epoch": 11.470235717139433,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 28710
    },
    {
      "epoch": 11.470635237714742,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 28711
    },
    {
      "epoch": 11.471034758290052,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 28712
    },
    {
      "epoch": 11.471434278865361,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 28713
    },
    {
      "epoch": 11.471833799440672,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 28714
    },
    {
      "epoch": 11.472233320015981,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 28715
    },
    {
      "epoch": 11.47263284059129,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 28716
    },
    {
      "epoch": 11.4730323611666,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 28717
    },
    {
      "epoch": 11.47343188174191,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 28718
    },
    {
      "epoch": 11.47383140231722,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 28719
    },
    {
      "epoch": 11.474230922892529,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 28720
    },
    {
      "epoch": 11.474630443467838,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 28721
    },
    {
      "epoch": 11.475029964043149,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 28722
    },
    {
      "epoch": 11.475429484618457,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 28723
    },
    {
      "epoch": 11.475829005193768,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 28724
    },
    {
      "epoch": 11.476228525769077,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 28725
    },
    {
      "epoch": 11.476628046344386,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 28726
    },
    {
      "epoch": 11.477027566919697,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 28727
    },
    {
      "epoch": 11.477427087495006,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 28728
    },
    {
      "epoch": 11.477826608070316,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 28729
    },
    {
      "epoch": 11.478226128645625,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 28730
    },
    {
      "epoch": 11.478625649220934,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 28731
    },
    {
      "epoch": 11.479025169796245,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 28732
    },
    {
      "epoch": 11.479424690371554,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 28733
    },
    {
      "epoch": 11.479824210946864,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 28734
    },
    {
      "epoch": 11.480223731522173,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 28735
    },
    {
      "epoch": 11.480623252097484,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 28736
    },
    {
      "epoch": 11.481022772672793,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 28737
    },
    {
      "epoch": 11.481422293248102,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 28738
    },
    {
      "epoch": 11.481821813823412,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 28739
    },
    {
      "epoch": 11.482221334398721,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 28740
    },
    {
      "epoch": 11.482620854974032,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 28741
    },
    {
      "epoch": 11.48302037554934,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 28742
    },
    {
      "epoch": 11.48341989612465,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 28743
    },
    {
      "epoch": 11.48381941669996,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 28744
    },
    {
      "epoch": 11.48421893727527,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 28745
    },
    {
      "epoch": 11.48461845785058,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 28746
    },
    {
      "epoch": 11.485017978425889,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 28747
    },
    {
      "epoch": 11.485417499001198,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 28748
    },
    {
      "epoch": 11.485817019576508,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 28749
    },
    {
      "epoch": 11.486216540151817,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 28750
    },
    {
      "epoch": 11.486616060727128,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 28751
    },
    {
      "epoch": 11.487015581302437,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 28752
    },
    {
      "epoch": 11.487415101877748,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 28753
    },
    {
      "epoch": 11.487814622453056,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 28754
    },
    {
      "epoch": 11.488214143028365,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 28755
    },
    {
      "epoch": 11.488613663603676,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 28756
    },
    {
      "epoch": 11.489013184178985,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 28757
    },
    {
      "epoch": 11.489412704754296,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 28758
    },
    {
      "epoch": 11.489812225329604,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 28759
    },
    {
      "epoch": 11.490211745904913,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 28760
    },
    {
      "epoch": 11.490611266480224,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 28761
    },
    {
      "epoch": 11.491010787055533,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 28762
    },
    {
      "epoch": 11.491410307630844,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 28763
    },
    {
      "epoch": 11.491809828206152,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 28764
    },
    {
      "epoch": 11.492209348781461,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 28765
    },
    {
      "epoch": 11.492608869356772,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 28766
    },
    {
      "epoch": 11.493008389932081,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 28767
    },
    {
      "epoch": 11.493407910507392,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 28768
    },
    {
      "epoch": 11.4938074310827,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 28769
    },
    {
      "epoch": 11.494206951658011,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 28770
    },
    {
      "epoch": 11.49460647223332,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 28771
    },
    {
      "epoch": 11.495005992808629,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 28772
    },
    {
      "epoch": 11.49540551338394,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 28773
    },
    {
      "epoch": 11.495805033959249,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 28774
    },
    {
      "epoch": 11.49620455453456,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 28775
    },
    {
      "epoch": 11.496604075109868,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 28776
    },
    {
      "epoch": 11.497003595685177,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 28777
    },
    {
      "epoch": 11.497403116260488,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 28778
    },
    {
      "epoch": 11.497802636835797,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 28779
    },
    {
      "epoch": 11.498202157411107,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 28780
    },
    {
      "epoch": 11.498601677986416,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 28781
    },
    {
      "epoch": 11.499001198561725,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 28782
    },
    {
      "epoch": 11.499400719137036,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 28783
    },
    {
      "epoch": 11.499800239712345,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 28784
    },
    {
      "epoch": 11.500199760287655,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 28785
    },
    {
      "epoch": 11.500599280862964,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 28786
    },
    {
      "epoch": 11.500998801438275,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 28787
    },
    {
      "epoch": 11.501398322013584,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 28788
    },
    {
      "epoch": 11.501797842588893,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 28789
    },
    {
      "epoch": 11.502197363164203,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 28790
    },
    {
      "epoch": 11.502596883739512,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 28791
    },
    {
      "epoch": 11.502996404314823,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 28792
    },
    {
      "epoch": 11.503395924890132,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 28793
    },
    {
      "epoch": 11.50379544546544,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 28794
    },
    {
      "epoch": 11.504194966040751,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 28795
    },
    {
      "epoch": 11.50459448661606,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 28796
    },
    {
      "epoch": 11.504994007191371,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 28797
    },
    {
      "epoch": 11.50539352776668,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 28798
    },
    {
      "epoch": 11.505793048341989,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 28799
    },
    {
      "epoch": 11.5061925689173,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 28800
    },
    {
      "epoch": 11.506592089492608,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 28801
    },
    {
      "epoch": 11.506991610067919,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 28802
    },
    {
      "epoch": 11.507391130643228,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 28803
    },
    {
      "epoch": 11.507790651218539,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 28804
    },
    {
      "epoch": 11.508190171793848,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 28805
    },
    {
      "epoch": 11.508589692369156,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 28806
    },
    {
      "epoch": 11.508989212944467,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 28807
    },
    {
      "epoch": 11.509388733519776,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 28808
    },
    {
      "epoch": 11.509788254095087,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 28809
    },
    {
      "epoch": 11.510187774670396,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 28810
    },
    {
      "epoch": 11.510587295245704,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 28811
    },
    {
      "epoch": 11.510986815821015,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 28812
    },
    {
      "epoch": 11.511386336396324,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 28813
    },
    {
      "epoch": 11.511785856971635,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 28814
    },
    {
      "epoch": 11.512185377546944,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 28815
    },
    {
      "epoch": 11.512584898122252,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 28816
    },
    {
      "epoch": 11.512984418697563,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 28817
    },
    {
      "epoch": 11.513383939272872,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 28818
    },
    {
      "epoch": 11.513783459848183,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 28819
    },
    {
      "epoch": 11.514182980423492,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 28820
    },
    {
      "epoch": 11.514582500998802,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 28821
    },
    {
      "epoch": 11.514982021574111,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 28822
    },
    {
      "epoch": 11.51538154214942,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 28823
    },
    {
      "epoch": 11.51578106272473,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 28824
    },
    {
      "epoch": 11.51618058330004,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 28825
    },
    {
      "epoch": 11.51658010387535,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 28826
    },
    {
      "epoch": 11.51697962445066,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 28827
    },
    {
      "epoch": 11.517379145025968,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 28828
    },
    {
      "epoch": 11.517778665601279,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 28829
    },
    {
      "epoch": 11.518178186176588,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 28830
    },
    {
      "epoch": 11.518577706751898,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 28831
    },
    {
      "epoch": 11.518977227327207,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 28832
    },
    {
      "epoch": 11.519376747902516,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 28833
    },
    {
      "epoch": 11.519776268477827,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 28834
    },
    {
      "epoch": 11.520175789053136,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 28835
    },
    {
      "epoch": 11.520575309628446,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 28836
    },
    {
      "epoch": 11.520974830203755,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 28837
    },
    {
      "epoch": 11.521374350779066,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 28838
    },
    {
      "epoch": 11.521773871354375,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 28839
    },
    {
      "epoch": 11.522173391929684,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 28840
    },
    {
      "epoch": 11.522572912504994,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 28841
    },
    {
      "epoch": 11.522972433080303,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 28842
    },
    {
      "epoch": 11.523371953655614,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 28843
    },
    {
      "epoch": 11.523771474230923,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 28844
    },
    {
      "epoch": 11.524170994806232,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 28845
    },
    {
      "epoch": 11.524570515381543,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 28846
    },
    {
      "epoch": 11.524970035956851,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 28847
    },
    {
      "epoch": 11.525369556532162,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 28848
    },
    {
      "epoch": 11.525769077107471,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 28849
    },
    {
      "epoch": 11.52616859768278,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 28850
    },
    {
      "epoch": 11.52656811825809,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 28851
    },
    {
      "epoch": 11.5269676388334,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 28852
    },
    {
      "epoch": 11.52736715940871,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0502,
      "step": 28853
    },
    {
      "epoch": 11.527766679984019,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 28854
    },
    {
      "epoch": 11.52816620055933,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 28855
    },
    {
      "epoch": 11.528565721134639,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 28856
    },
    {
      "epoch": 11.528965241709948,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 28857
    },
    {
      "epoch": 11.529364762285258,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 28858
    },
    {
      "epoch": 11.529764282860567,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 28859
    },
    {
      "epoch": 11.530163803435878,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 28860
    },
    {
      "epoch": 11.530563324011187,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 28861
    },
    {
      "epoch": 11.530962844586496,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 28862
    },
    {
      "epoch": 11.531362365161806,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 28863
    },
    {
      "epoch": 11.531761885737115,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 28864
    },
    {
      "epoch": 11.532161406312426,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 28865
    },
    {
      "epoch": 11.532560926887735,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 28866
    },
    {
      "epoch": 11.532960447463044,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 28867
    },
    {
      "epoch": 11.533359968038354,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 28868
    },
    {
      "epoch": 11.533759488613663,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 28869
    },
    {
      "epoch": 11.534159009188974,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 28870
    },
    {
      "epoch": 11.534558529764283,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 28871
    },
    {
      "epoch": 11.534958050339593,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.012,
      "step": 28872
    },
    {
      "epoch": 11.535357570914902,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 28873
    },
    {
      "epoch": 11.535757091490211,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 28874
    },
    {
      "epoch": 11.536156612065522,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 28875
    },
    {
      "epoch": 11.53655613264083,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 28876
    },
    {
      "epoch": 11.53695565321614,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 28877
    },
    {
      "epoch": 11.53735517379145,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 28878
    },
    {
      "epoch": 11.53775469436676,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 28879
    },
    {
      "epoch": 11.53815421494207,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 28880
    },
    {
      "epoch": 11.538553735517379,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 28881
    },
    {
      "epoch": 11.53895325609269,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 28882
    },
    {
      "epoch": 11.539352776667998,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 28883
    },
    {
      "epoch": 11.539752297243307,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 28884
    },
    {
      "epoch": 11.540151817818618,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 28885
    },
    {
      "epoch": 11.540551338393927,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 28886
    },
    {
      "epoch": 11.540950858969238,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 28887
    },
    {
      "epoch": 11.541350379544546,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 28888
    },
    {
      "epoch": 11.541749900119855,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 28889
    },
    {
      "epoch": 11.542149420695166,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 28890
    },
    {
      "epoch": 11.542548941270475,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 28891
    },
    {
      "epoch": 11.542948461845786,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 28892
    },
    {
      "epoch": 11.543347982421094,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 28893
    },
    {
      "epoch": 11.543747502996403,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 28894
    },
    {
      "epoch": 11.544147023571714,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 28895
    },
    {
      "epoch": 11.544546544147023,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 28896
    },
    {
      "epoch": 11.544946064722334,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 28897
    },
    {
      "epoch": 11.545345585297643,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 28898
    },
    {
      "epoch": 11.545745105872953,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 28899
    },
    {
      "epoch": 11.546144626448262,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 28900
    },
    {
      "epoch": 11.546544147023571,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 28901
    },
    {
      "epoch": 11.546943667598882,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 28902
    },
    {
      "epoch": 11.54734318817419,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 28903
    },
    {
      "epoch": 11.547742708749501,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 28904
    },
    {
      "epoch": 11.54814222932481,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 28905
    },
    {
      "epoch": 11.548541749900119,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 28906
    },
    {
      "epoch": 11.54894127047543,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 28907
    },
    {
      "epoch": 11.549340791050739,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 28908
    },
    {
      "epoch": 11.54974031162605,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 28909
    },
    {
      "epoch": 11.550139832201358,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 28910
    },
    {
      "epoch": 11.550539352776667,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 28911
    },
    {
      "epoch": 11.550938873351978,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 28912
    },
    {
      "epoch": 11.551338393927287,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 28913
    },
    {
      "epoch": 11.551737914502597,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 28914
    },
    {
      "epoch": 11.552137435077906,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 28915
    },
    {
      "epoch": 11.552536955653217,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 28916
    },
    {
      "epoch": 11.552936476228526,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 28917
    },
    {
      "epoch": 11.553335996803835,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 28918
    },
    {
      "epoch": 11.553735517379145,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 28919
    },
    {
      "epoch": 11.554135037954454,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 28920
    },
    {
      "epoch": 11.554534558529765,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 28921
    },
    {
      "epoch": 11.554934079105074,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 28922
    },
    {
      "epoch": 11.555333599680383,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 28923
    },
    {
      "epoch": 11.555733120255693,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 28924
    },
    {
      "epoch": 11.556132640831002,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 28925
    },
    {
      "epoch": 11.556532161406313,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 28926
    },
    {
      "epoch": 11.556931681981622,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 28927
    },
    {
      "epoch": 11.55733120255693,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 28928
    },
    {
      "epoch": 11.557730723132241,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 28929
    },
    {
      "epoch": 11.55813024370755,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 28930
    },
    {
      "epoch": 11.558529764282861,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 28931
    },
    {
      "epoch": 11.55892928485817,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 28932
    },
    {
      "epoch": 11.55932880543348,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 28933
    },
    {
      "epoch": 11.55972832600879,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 28934
    },
    {
      "epoch": 11.560127846584098,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 28935
    },
    {
      "epoch": 11.560527367159409,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 28936
    },
    {
      "epoch": 11.560926887734718,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 28937
    },
    {
      "epoch": 11.561326408310029,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 28938
    },
    {
      "epoch": 11.561725928885338,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 28939
    },
    {
      "epoch": 11.562125449460646,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 28940
    },
    {
      "epoch": 11.562524970035957,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 28941
    },
    {
      "epoch": 11.562924490611266,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 28942
    },
    {
      "epoch": 11.563324011186577,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 28943
    },
    {
      "epoch": 11.563723531761886,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 28944
    },
    {
      "epoch": 11.564123052337195,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 28945
    },
    {
      "epoch": 11.564522572912505,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 28946
    },
    {
      "epoch": 11.564922093487814,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 28947
    },
    {
      "epoch": 11.565321614063125,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 28948
    },
    {
      "epoch": 11.565721134638434,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 28949
    },
    {
      "epoch": 11.566120655213744,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 28950
    },
    {
      "epoch": 11.566520175789053,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 28951
    },
    {
      "epoch": 11.566919696364362,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 28952
    },
    {
      "epoch": 11.567319216939673,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 28953
    },
    {
      "epoch": 11.567718737514982,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 28954
    },
    {
      "epoch": 11.568118258090292,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 28955
    },
    {
      "epoch": 11.568517778665601,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 28956
    },
    {
      "epoch": 11.56891729924091,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 28957
    },
    {
      "epoch": 11.56931681981622,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 28958
    },
    {
      "epoch": 11.56971634039153,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 28959
    },
    {
      "epoch": 11.57011586096684,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 28960
    },
    {
      "epoch": 11.57051538154215,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 28961
    },
    {
      "epoch": 11.570914902117458,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 28962
    },
    {
      "epoch": 11.571314422692769,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 28963
    },
    {
      "epoch": 11.571713943268078,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 28964
    },
    {
      "epoch": 11.572113463843388,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 28965
    },
    {
      "epoch": 11.572512984418697,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 28966
    },
    {
      "epoch": 11.572912504994008,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 28967
    },
    {
      "epoch": 11.573312025569317,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 28968
    },
    {
      "epoch": 11.573711546144626,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 28969
    },
    {
      "epoch": 11.574111066719937,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 28970
    },
    {
      "epoch": 11.574510587295245,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 28971
    },
    {
      "epoch": 11.574910107870556,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 28972
    },
    {
      "epoch": 11.575309628445865,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 28973
    },
    {
      "epoch": 11.575709149021174,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 28974
    },
    {
      "epoch": 11.576108669596485,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 28975
    },
    {
      "epoch": 11.576508190171793,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 28976
    },
    {
      "epoch": 11.576907710747104,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 28977
    },
    {
      "epoch": 11.577307231322413,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 28978
    },
    {
      "epoch": 11.577706751897722,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 28979
    },
    {
      "epoch": 11.578106272473033,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 28980
    },
    {
      "epoch": 11.578505793048341,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 28981
    },
    {
      "epoch": 11.578905313623652,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 28982
    },
    {
      "epoch": 11.579304834198961,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 28983
    },
    {
      "epoch": 11.579704354774272,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 28984
    },
    {
      "epoch": 11.58010387534958,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 28985
    },
    {
      "epoch": 11.58050339592489,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 28986
    },
    {
      "epoch": 11.5809029165002,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 28987
    },
    {
      "epoch": 11.581302437075509,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 28988
    },
    {
      "epoch": 11.58170195765082,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 28989
    },
    {
      "epoch": 11.582101478226129,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 28990
    },
    {
      "epoch": 11.582500998801438,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 28991
    },
    {
      "epoch": 11.582900519376748,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 28992
    },
    {
      "epoch": 11.583300039952057,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 28993
    },
    {
      "epoch": 11.583699560527368,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 28994
    },
    {
      "epoch": 11.584099081102677,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 28995
    },
    {
      "epoch": 11.584498601677986,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 28996
    },
    {
      "epoch": 11.584898122253296,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 28997
    },
    {
      "epoch": 11.585297642828605,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 28998
    },
    {
      "epoch": 11.585697163403916,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 28999
    },
    {
      "epoch": 11.586096683979225,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 29000
    },
    {
      "epoch": 11.586496204554535,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 29001
    },
    {
      "epoch": 11.586895725129844,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 29002
    },
    {
      "epoch": 11.587295245705153,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 29003
    },
    {
      "epoch": 11.587694766280464,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 29004
    },
    {
      "epoch": 11.588094286855773,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 29005
    },
    {
      "epoch": 11.588493807431083,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 29006
    },
    {
      "epoch": 11.588893328006392,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 29007
    },
    {
      "epoch": 11.589292848581701,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 29008
    },
    {
      "epoch": 11.589692369157012,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 29009
    },
    {
      "epoch": 11.59009188973232,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 29010
    },
    {
      "epoch": 11.590491410307632,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 29011
    },
    {
      "epoch": 11.59089093088294,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 29012
    },
    {
      "epoch": 11.59129045145825,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 29013
    },
    {
      "epoch": 11.59168997203356,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 29014
    },
    {
      "epoch": 11.592089492608869,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 29015
    },
    {
      "epoch": 11.59248901318418,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 29016
    },
    {
      "epoch": 11.592888533759488,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 29017
    },
    {
      "epoch": 11.5932880543348,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 29018
    },
    {
      "epoch": 11.593687574910108,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 29019
    },
    {
      "epoch": 11.594087095485417,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 29020
    },
    {
      "epoch": 11.594486616060728,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 29021
    },
    {
      "epoch": 11.594886136636037,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 29022
    },
    {
      "epoch": 11.595285657211347,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 29023
    },
    {
      "epoch": 11.595685177786656,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 29024
    },
    {
      "epoch": 11.596084698361965,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 29025
    },
    {
      "epoch": 11.596484218937276,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 29026
    },
    {
      "epoch": 11.596883739512585,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 29027
    },
    {
      "epoch": 11.597283260087895,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 29028
    },
    {
      "epoch": 11.597682780663204,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 29029
    },
    {
      "epoch": 11.598082301238513,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 29030
    },
    {
      "epoch": 11.598481821813824,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 29031
    },
    {
      "epoch": 11.598881342389133,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 29032
    },
    {
      "epoch": 11.599280862964443,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 29033
    },
    {
      "epoch": 11.599680383539752,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 29034
    },
    {
      "epoch": 11.600079904115063,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 29035
    },
    {
      "epoch": 11.600479424690372,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 29036
    },
    {
      "epoch": 11.60087894526568,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 29037
    },
    {
      "epoch": 11.601278465840991,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 29038
    },
    {
      "epoch": 11.6016779864163,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 29039
    },
    {
      "epoch": 11.602077506991611,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 29040
    },
    {
      "epoch": 11.60247702756692,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 29041
    },
    {
      "epoch": 11.602876548142229,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 29042
    },
    {
      "epoch": 11.60327606871754,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 29043
    },
    {
      "epoch": 11.603675589292848,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 29044
    },
    {
      "epoch": 11.604075109868159,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 29045
    },
    {
      "epoch": 11.604474630443468,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 29046
    },
    {
      "epoch": 11.604874151018777,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 29047
    },
    {
      "epoch": 11.605273671594087,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 29048
    },
    {
      "epoch": 11.605673192169396,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 29049
    },
    {
      "epoch": 11.606072712744707,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 29050
    },
    {
      "epoch": 11.606472233320016,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 29051
    },
    {
      "epoch": 11.606871753895327,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 29052
    },
    {
      "epoch": 11.607271274470635,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 29053
    },
    {
      "epoch": 11.607670795045944,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 29054
    },
    {
      "epoch": 11.608070315621255,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 29055
    },
    {
      "epoch": 11.608469836196564,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 29056
    },
    {
      "epoch": 11.608869356771873,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 29057
    },
    {
      "epoch": 11.609268877347183,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 29058
    },
    {
      "epoch": 11.609668397922492,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 29059
    },
    {
      "epoch": 11.610067918497803,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 29060
    },
    {
      "epoch": 11.610467439073112,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 29061
    },
    {
      "epoch": 11.610866959648423,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 29062
    },
    {
      "epoch": 11.611266480223732,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 29063
    },
    {
      "epoch": 11.61166600079904,
      "grad_norm": 0.036376953125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 29064
    },
    {
      "epoch": 11.612065521374351,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 29065
    },
    {
      "epoch": 11.61246504194966,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 29066
    },
    {
      "epoch": 11.61286456252497,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 29067
    },
    {
      "epoch": 11.61326408310028,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 29068
    },
    {
      "epoch": 11.613663603675588,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 29069
    },
    {
      "epoch": 11.6140631242509,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 29070
    },
    {
      "epoch": 11.614462644826208,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 29071
    },
    {
      "epoch": 11.614862165401519,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 29072
    },
    {
      "epoch": 11.615261685976828,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 29073
    },
    {
      "epoch": 11.615661206552137,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 29074
    },
    {
      "epoch": 11.616060727127447,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0112,
      "step": 29075
    },
    {
      "epoch": 11.616460247702756,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 29076
    },
    {
      "epoch": 11.616859768278067,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 29077
    },
    {
      "epoch": 11.617259288853376,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0473,
      "step": 29078
    },
    {
      "epoch": 11.617658809428686,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 29079
    },
    {
      "epoch": 11.618058330003995,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 29080
    },
    {
      "epoch": 11.618457850579304,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 29081
    },
    {
      "epoch": 11.618857371154615,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 29082
    },
    {
      "epoch": 11.619256891729924,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 29083
    },
    {
      "epoch": 11.619656412305234,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 29084
    },
    {
      "epoch": 11.620055932880543,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 29085
    },
    {
      "epoch": 11.620455453455852,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0505,
      "step": 29086
    },
    {
      "epoch": 11.620854974031163,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 29087
    },
    {
      "epoch": 11.621254494606472,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 29088
    },
    {
      "epoch": 11.621654015181782,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 29089
    },
    {
      "epoch": 11.622053535757091,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 29090
    },
    {
      "epoch": 11.6224530563324,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 29091
    },
    {
      "epoch": 11.622852576907711,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 29092
    },
    {
      "epoch": 11.62325209748302,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 29093
    },
    {
      "epoch": 11.62365161805833,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 29094
    },
    {
      "epoch": 11.62405113863364,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 29095
    },
    {
      "epoch": 11.62445065920895,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 29096
    },
    {
      "epoch": 11.624850179784259,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 29097
    },
    {
      "epoch": 11.625249700359568,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 29098
    },
    {
      "epoch": 11.625649220934879,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0504,
      "step": 29099
    },
    {
      "epoch": 11.626048741510187,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 29100
    },
    {
      "epoch": 11.626448262085498,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 29101
    },
    {
      "epoch": 11.626847782660807,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 29102
    },
    {
      "epoch": 11.627247303236116,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 29103
    },
    {
      "epoch": 11.627646823811427,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 29104
    },
    {
      "epoch": 11.628046344386735,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 29105
    },
    {
      "epoch": 11.628445864962046,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 29106
    },
    {
      "epoch": 11.628845385537355,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 29107
    },
    {
      "epoch": 11.629244906112664,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 29108
    },
    {
      "epoch": 11.629644426687975,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 29109
    },
    {
      "epoch": 11.630043947263283,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 29110
    },
    {
      "epoch": 11.630443467838594,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 29111
    },
    {
      "epoch": 11.630842988413903,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 29112
    },
    {
      "epoch": 11.631242508989214,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 29113
    },
    {
      "epoch": 11.631642029564523,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 29114
    },
    {
      "epoch": 11.632041550139832,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 29115
    },
    {
      "epoch": 11.632441070715142,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 29116
    },
    {
      "epoch": 11.632840591290451,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 29117
    },
    {
      "epoch": 11.633240111865762,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 29118
    },
    {
      "epoch": 11.63363963244107,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 29119
    },
    {
      "epoch": 11.63403915301638,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 29120
    },
    {
      "epoch": 11.63443867359169,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 29121
    },
    {
      "epoch": 11.634838194167,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 29122
    },
    {
      "epoch": 11.63523771474231,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 29123
    },
    {
      "epoch": 11.635637235317619,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 29124
    },
    {
      "epoch": 11.636036755892928,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 29125
    },
    {
      "epoch": 11.636436276468238,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 29126
    },
    {
      "epoch": 11.636835797043547,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 29127
    },
    {
      "epoch": 11.637235317618858,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 29128
    },
    {
      "epoch": 11.637634838194167,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 29129
    },
    {
      "epoch": 11.638034358769477,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 29130
    },
    {
      "epoch": 11.638433879344786,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 29131
    },
    {
      "epoch": 11.638833399920095,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 29132
    },
    {
      "epoch": 11.639232920495406,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 29133
    },
    {
      "epoch": 11.639632441070715,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 29134
    },
    {
      "epoch": 11.640031961646025,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 29135
    },
    {
      "epoch": 11.640431482221334,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 29136
    },
    {
      "epoch": 11.640831002796643,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 29137
    },
    {
      "epoch": 11.641230523371954,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 29138
    },
    {
      "epoch": 11.641630043947263,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 29139
    },
    {
      "epoch": 11.642029564522574,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 29140
    },
    {
      "epoch": 11.642429085097882,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 29141
    },
    {
      "epoch": 11.642828605673191,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 29142
    },
    {
      "epoch": 11.643228126248502,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 29143
    },
    {
      "epoch": 11.643627646823811,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 29144
    },
    {
      "epoch": 11.644027167399122,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 29145
    },
    {
      "epoch": 11.64442668797443,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 29146
    },
    {
      "epoch": 11.644826208549741,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 29147
    },
    {
      "epoch": 11.64522572912505,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 29148
    },
    {
      "epoch": 11.645625249700359,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 29149
    },
    {
      "epoch": 11.64602477027567,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 29150
    },
    {
      "epoch": 11.646424290850979,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 29151
    },
    {
      "epoch": 11.64682381142629,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 29152
    },
    {
      "epoch": 11.647223332001598,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 29153
    },
    {
      "epoch": 11.647622852576907,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 29154
    },
    {
      "epoch": 11.648022373152218,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 29155
    },
    {
      "epoch": 11.648421893727527,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 29156
    },
    {
      "epoch": 11.648821414302837,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 29157
    },
    {
      "epoch": 11.649220934878146,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 29158
    },
    {
      "epoch": 11.649620455453455,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 29159
    },
    {
      "epoch": 11.650019976028766,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 29160
    },
    {
      "epoch": 11.650419496604075,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 29161
    },
    {
      "epoch": 11.650819017179385,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 29162
    },
    {
      "epoch": 11.651218537754694,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 29163
    },
    {
      "epoch": 11.651618058330005,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 29164
    },
    {
      "epoch": 11.652017578905314,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 29165
    },
    {
      "epoch": 11.652417099480623,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 29166
    },
    {
      "epoch": 11.652816620055933,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 29167
    },
    {
      "epoch": 11.653216140631242,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 29168
    },
    {
      "epoch": 11.653615661206553,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 29169
    },
    {
      "epoch": 11.654015181781862,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 29170
    },
    {
      "epoch": 11.65441470235717,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 29171
    },
    {
      "epoch": 11.654814222932481,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 29172
    },
    {
      "epoch": 11.65521374350779,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 29173
    },
    {
      "epoch": 11.655613264083101,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 29174
    },
    {
      "epoch": 11.65601278465841,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 29175
    },
    {
      "epoch": 11.656412305233719,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 29176
    },
    {
      "epoch": 11.65681182580903,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 29177
    },
    {
      "epoch": 11.657211346384338,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 29178
    },
    {
      "epoch": 11.657610866959649,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 29179
    },
    {
      "epoch": 11.658010387534958,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 29180
    },
    {
      "epoch": 11.658409908110269,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 29181
    },
    {
      "epoch": 11.658809428685577,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 29182
    },
    {
      "epoch": 11.659208949260886,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 29183
    },
    {
      "epoch": 11.659608469836197,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 29184
    },
    {
      "epoch": 11.660007990411506,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 29185
    },
    {
      "epoch": 11.660407510986817,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 29186
    },
    {
      "epoch": 11.660807031562126,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 29187
    },
    {
      "epoch": 11.661206552137434,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0506,
      "step": 29188
    },
    {
      "epoch": 11.661606072712745,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 29189
    },
    {
      "epoch": 11.662005593288054,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 29190
    },
    {
      "epoch": 11.662405113863365,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 29191
    },
    {
      "epoch": 11.662804634438674,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 29192
    },
    {
      "epoch": 11.663204155013982,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 29193
    },
    {
      "epoch": 11.663603675589293,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 29194
    },
    {
      "epoch": 11.664003196164602,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 29195
    },
    {
      "epoch": 11.664402716739913,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 29196
    },
    {
      "epoch": 11.664802237315222,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 29197
    },
    {
      "epoch": 11.665201757890532,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 29198
    },
    {
      "epoch": 11.665601278465841,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 29199
    },
    {
      "epoch": 11.66600079904115,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 29200
    },
    {
      "epoch": 11.66640031961646,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0475,
      "step": 29201
    },
    {
      "epoch": 11.66679984019177,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 29202
    },
    {
      "epoch": 11.66719936076708,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 29203
    },
    {
      "epoch": 11.66759888134239,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 29204
    },
    {
      "epoch": 11.667998401917698,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 29205
    },
    {
      "epoch": 11.668397922493009,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 29206
    },
    {
      "epoch": 11.668797443068318,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 29207
    },
    {
      "epoch": 11.669196963643628,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 29208
    },
    {
      "epoch": 11.669596484218937,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 29209
    },
    {
      "epoch": 11.669996004794246,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 29210
    },
    {
      "epoch": 11.670395525369557,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 29211
    },
    {
      "epoch": 11.670795045944866,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 29212
    },
    {
      "epoch": 11.671194566520176,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 29213
    },
    {
      "epoch": 11.671594087095485,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 29214
    },
    {
      "epoch": 11.671993607670796,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 29215
    },
    {
      "epoch": 11.672393128246105,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 29216
    },
    {
      "epoch": 11.672792648821414,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 29217
    },
    {
      "epoch": 11.673192169396724,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 29218
    },
    {
      "epoch": 11.673591689972033,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 29219
    },
    {
      "epoch": 11.673991210547344,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 29220
    },
    {
      "epoch": 11.674390731122653,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 29221
    },
    {
      "epoch": 11.674790251697962,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 29222
    },
    {
      "epoch": 11.675189772273272,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 29223
    },
    {
      "epoch": 11.675589292848581,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 29224
    },
    {
      "epoch": 11.675988813423892,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 29225
    },
    {
      "epoch": 11.676388333999201,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 29226
    },
    {
      "epoch": 11.67678785457451,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 29227
    },
    {
      "epoch": 11.67718737514982,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 29228
    },
    {
      "epoch": 11.67758689572513,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 29229
    },
    {
      "epoch": 11.67798641630044,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 29230
    },
    {
      "epoch": 11.678385936875749,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 29231
    },
    {
      "epoch": 11.67878545745106,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 29232
    },
    {
      "epoch": 11.679184978026369,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 29233
    },
    {
      "epoch": 11.679584498601677,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 29234
    },
    {
      "epoch": 11.679984019176988,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 29235
    },
    {
      "epoch": 11.680383539752297,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 29236
    },
    {
      "epoch": 11.680783060327606,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 29237
    },
    {
      "epoch": 11.681182580902917,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 29238
    },
    {
      "epoch": 11.681582101478226,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 29239
    },
    {
      "epoch": 11.681981622053536,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 29240
    },
    {
      "epoch": 11.682381142628845,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 29241
    },
    {
      "epoch": 11.682780663204156,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 29242
    },
    {
      "epoch": 11.683180183779465,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 29243
    },
    {
      "epoch": 11.683579704354774,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 29244
    },
    {
      "epoch": 11.683979224930084,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 29245
    },
    {
      "epoch": 11.684378745505393,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 29246
    },
    {
      "epoch": 11.684778266080704,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 29247
    },
    {
      "epoch": 11.685177786656013,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 29248
    },
    {
      "epoch": 11.685577307231322,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 29249
    },
    {
      "epoch": 11.685976827806632,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 29250
    },
    {
      "epoch": 11.686376348381941,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 29251
    },
    {
      "epoch": 11.686775868957252,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 29252
    },
    {
      "epoch": 11.68717538953256,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 29253
    },
    {
      "epoch": 11.68757491010787,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 29254
    },
    {
      "epoch": 11.68797443068318,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 29255
    },
    {
      "epoch": 11.68837395125849,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 29256
    },
    {
      "epoch": 11.6887734718338,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 29257
    },
    {
      "epoch": 11.689172992409109,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 29258
    },
    {
      "epoch": 11.68957251298442,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 29259
    },
    {
      "epoch": 11.689972033559728,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 29260
    },
    {
      "epoch": 11.690371554135037,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 29261
    },
    {
      "epoch": 11.690771074710348,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 29262
    },
    {
      "epoch": 11.691170595285657,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 29263
    },
    {
      "epoch": 11.691570115860968,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 29264
    },
    {
      "epoch": 11.691969636436276,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 29265
    },
    {
      "epoch": 11.692369157011585,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 29266
    },
    {
      "epoch": 11.692768677586896,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 29267
    },
    {
      "epoch": 11.693168198162205,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 29268
    },
    {
      "epoch": 11.693567718737516,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 29269
    },
    {
      "epoch": 11.693967239312824,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 29270
    },
    {
      "epoch": 11.694366759888133,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 29271
    },
    {
      "epoch": 11.694766280463444,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 29272
    },
    {
      "epoch": 11.695165801038753,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 29273
    },
    {
      "epoch": 11.695565321614064,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 29274
    },
    {
      "epoch": 11.695964842189372,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 29275
    },
    {
      "epoch": 11.696364362764683,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 29276
    },
    {
      "epoch": 11.696763883339992,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 29277
    },
    {
      "epoch": 11.697163403915301,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 29278
    },
    {
      "epoch": 11.697562924490612,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 29279
    },
    {
      "epoch": 11.69796244506592,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 29280
    },
    {
      "epoch": 11.698361965641231,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 29281
    },
    {
      "epoch": 11.69876148621654,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 29282
    },
    {
      "epoch": 11.699161006791849,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 29283
    },
    {
      "epoch": 11.69956052736716,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 29284
    },
    {
      "epoch": 11.699960047942469,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 29285
    },
    {
      "epoch": 11.70035956851778,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 29286
    },
    {
      "epoch": 11.700759089093088,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 29287
    },
    {
      "epoch": 11.701158609668397,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 29288
    },
    {
      "epoch": 11.701558130243708,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 29289
    },
    {
      "epoch": 11.701957650819017,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 29290
    },
    {
      "epoch": 11.702357171394327,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 29291
    },
    {
      "epoch": 11.702756691969636,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 29292
    },
    {
      "epoch": 11.703156212544947,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 29293
    },
    {
      "epoch": 11.703555733120256,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 29294
    },
    {
      "epoch": 11.703955253695565,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 29295
    },
    {
      "epoch": 11.704354774270875,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 29296
    },
    {
      "epoch": 11.704754294846184,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 29297
    },
    {
      "epoch": 11.705153815421495,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 29298
    },
    {
      "epoch": 11.705553335996804,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 29299
    },
    {
      "epoch": 11.705952856572113,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 29300
    },
    {
      "epoch": 11.706352377147423,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 29301
    },
    {
      "epoch": 11.706751897722732,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 29302
    },
    {
      "epoch": 11.707151418298043,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 29303
    },
    {
      "epoch": 11.707550938873352,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 29304
    },
    {
      "epoch": 11.70795045944866,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 29305
    },
    {
      "epoch": 11.708349980023971,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 29306
    },
    {
      "epoch": 11.70874950059928,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 29307
    },
    {
      "epoch": 11.709149021174591,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 29308
    },
    {
      "epoch": 11.7095485417499,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 29309
    },
    {
      "epoch": 11.70994806232521,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 29310
    },
    {
      "epoch": 11.71034758290052,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 29311
    },
    {
      "epoch": 11.710747103475828,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 29312
    },
    {
      "epoch": 11.711146624051139,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 29313
    },
    {
      "epoch": 11.711546144626448,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 29314
    },
    {
      "epoch": 11.711945665201759,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 29315
    },
    {
      "epoch": 11.712345185777068,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 29316
    },
    {
      "epoch": 11.712744706352376,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 29317
    },
    {
      "epoch": 11.713144226927687,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 29318
    },
    {
      "epoch": 11.713543747502996,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 29319
    },
    {
      "epoch": 11.713943268078307,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 29320
    },
    {
      "epoch": 11.714342788653616,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 29321
    },
    {
      "epoch": 11.714742309228924,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 29322
    },
    {
      "epoch": 11.715141829804235,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 29323
    },
    {
      "epoch": 11.715541350379544,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 29324
    },
    {
      "epoch": 11.715940870954855,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 29325
    },
    {
      "epoch": 11.716340391530164,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 29326
    },
    {
      "epoch": 11.716739912105474,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 29327
    },
    {
      "epoch": 11.717139432680783,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 29328
    },
    {
      "epoch": 11.717538953256092,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 29329
    },
    {
      "epoch": 11.717938473831403,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 29330
    },
    {
      "epoch": 11.718337994406712,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 29331
    },
    {
      "epoch": 11.718737514982022,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 29332
    },
    {
      "epoch": 11.719137035557331,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 29333
    },
    {
      "epoch": 11.71953655613264,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 29334
    },
    {
      "epoch": 11.71993607670795,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 29335
    },
    {
      "epoch": 11.72033559728326,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 29336
    },
    {
      "epoch": 11.72073511785857,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 29337
    },
    {
      "epoch": 11.72113463843388,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 29338
    },
    {
      "epoch": 11.721534159009188,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 29339
    },
    {
      "epoch": 11.721933679584499,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 29340
    },
    {
      "epoch": 11.722333200159808,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 29341
    },
    {
      "epoch": 11.722732720735118,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 29342
    },
    {
      "epoch": 11.723132241310427,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 29343
    },
    {
      "epoch": 11.723531761885738,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 29344
    },
    {
      "epoch": 11.723931282461047,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 29345
    },
    {
      "epoch": 11.724330803036356,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 29346
    },
    {
      "epoch": 11.724730323611666,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 29347
    },
    {
      "epoch": 11.725129844186975,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 29348
    },
    {
      "epoch": 11.725529364762286,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 29349
    },
    {
      "epoch": 11.725928885337595,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 29350
    },
    {
      "epoch": 11.726328405912904,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 29351
    },
    {
      "epoch": 11.726727926488214,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 29352
    },
    {
      "epoch": 11.727127447063523,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 29353
    },
    {
      "epoch": 11.727526967638834,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 29354
    },
    {
      "epoch": 11.727926488214143,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 29355
    },
    {
      "epoch": 11.728326008789452,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 29356
    },
    {
      "epoch": 11.728725529364763,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 29357
    },
    {
      "epoch": 11.729125049940071,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 29358
    },
    {
      "epoch": 11.729524570515382,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 29359
    },
    {
      "epoch": 11.729924091090691,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 29360
    },
    {
      "epoch": 11.730323611666002,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 29361
    },
    {
      "epoch": 11.73072313224131,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 29362
    },
    {
      "epoch": 11.73112265281662,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 29363
    },
    {
      "epoch": 11.73152217339193,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 29364
    },
    {
      "epoch": 11.731921693967239,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 29365
    },
    {
      "epoch": 11.73232121454255,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 29366
    },
    {
      "epoch": 11.732720735117859,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 29367
    },
    {
      "epoch": 11.733120255693168,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 29368
    },
    {
      "epoch": 11.733519776268478,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 29369
    },
    {
      "epoch": 11.733919296843787,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 29370
    },
    {
      "epoch": 11.734318817419098,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 29371
    },
    {
      "epoch": 11.734718337994407,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 29372
    },
    {
      "epoch": 11.735117858569716,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 29373
    },
    {
      "epoch": 11.735517379145026,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 29374
    },
    {
      "epoch": 11.735916899720335,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 29375
    },
    {
      "epoch": 11.736316420295646,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 29376
    },
    {
      "epoch": 11.736715940870955,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 29377
    },
    {
      "epoch": 11.737115461446265,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 29378
    },
    {
      "epoch": 11.737514982021574,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 29379
    },
    {
      "epoch": 11.737914502596883,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 29380
    },
    {
      "epoch": 11.738314023172194,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 29381
    },
    {
      "epoch": 11.738713543747503,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 29382
    },
    {
      "epoch": 11.739113064322813,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 29383
    },
    {
      "epoch": 11.739512584898122,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 29384
    },
    {
      "epoch": 11.739912105473431,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 29385
    },
    {
      "epoch": 11.740311626048742,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 29386
    },
    {
      "epoch": 11.74071114662405,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 29387
    },
    {
      "epoch": 11.741110667199361,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 29388
    },
    {
      "epoch": 11.74151018777467,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 29389
    },
    {
      "epoch": 11.74190970834998,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 29390
    },
    {
      "epoch": 11.74230922892529,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 29391
    },
    {
      "epoch": 11.742708749500599,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 29392
    },
    {
      "epoch": 11.74310827007591,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 29393
    },
    {
      "epoch": 11.743507790651218,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 29394
    },
    {
      "epoch": 11.743907311226529,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 29395
    },
    {
      "epoch": 11.744306831801838,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 29396
    },
    {
      "epoch": 11.744706352377147,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 29397
    },
    {
      "epoch": 11.745105872952458,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 29398
    },
    {
      "epoch": 11.745505393527766,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 29399
    },
    {
      "epoch": 11.745904914103077,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 29400
    },
    {
      "epoch": 11.746304434678386,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 29401
    },
    {
      "epoch": 11.746703955253695,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 29402
    },
    {
      "epoch": 11.747103475829006,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 29403
    },
    {
      "epoch": 11.747502996404315,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 29404
    },
    {
      "epoch": 11.747902516979625,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 29405
    },
    {
      "epoch": 11.748302037554934,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 29406
    },
    {
      "epoch": 11.748701558130243,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 29407
    },
    {
      "epoch": 11.749101078705554,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 29408
    },
    {
      "epoch": 11.749500599280863,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 29409
    },
    {
      "epoch": 11.749900119856173,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 29410
    },
    {
      "epoch": 11.750299640431482,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 29411
    },
    {
      "epoch": 11.750699161006793,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 29412
    },
    {
      "epoch": 11.751098681582102,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 29413
    },
    {
      "epoch": 11.75149820215741,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 29414
    },
    {
      "epoch": 11.751897722732721,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 29415
    },
    {
      "epoch": 11.75229724330803,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 29416
    },
    {
      "epoch": 11.752696763883339,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 29417
    },
    {
      "epoch": 11.75309628445865,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 29418
    },
    {
      "epoch": 11.753495805033959,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 29419
    },
    {
      "epoch": 11.75389532560927,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 29420
    },
    {
      "epoch": 11.754294846184578,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 29421
    },
    {
      "epoch": 11.754694366759889,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 29422
    },
    {
      "epoch": 11.755093887335198,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 29423
    },
    {
      "epoch": 11.755493407910507,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 29424
    },
    {
      "epoch": 11.755892928485817,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 29425
    },
    {
      "epoch": 11.756292449061126,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 29426
    },
    {
      "epoch": 11.756691969636437,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 29427
    },
    {
      "epoch": 11.757091490211746,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 29428
    },
    {
      "epoch": 11.757491010787055,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 29429
    },
    {
      "epoch": 11.757890531362365,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 29430
    },
    {
      "epoch": 11.758290051937674,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 29431
    },
    {
      "epoch": 11.758689572512985,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 29432
    },
    {
      "epoch": 11.759089093088294,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 29433
    },
    {
      "epoch": 11.759488613663603,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 29434
    },
    {
      "epoch": 11.759888134238913,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 29435
    },
    {
      "epoch": 11.760287654814222,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 29436
    },
    {
      "epoch": 11.760687175389533,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 29437
    },
    {
      "epoch": 11.761086695964842,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 29438
    },
    {
      "epoch": 11.761486216540153,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 29439
    },
    {
      "epoch": 11.761885737115461,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 29440
    },
    {
      "epoch": 11.76228525769077,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 29441
    },
    {
      "epoch": 11.762684778266081,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 29442
    },
    {
      "epoch": 11.76308429884139,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 29443
    },
    {
      "epoch": 11.7634838194167,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 29444
    },
    {
      "epoch": 11.76388333999201,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 29445
    },
    {
      "epoch": 11.764282860567318,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 29446
    },
    {
      "epoch": 11.764682381142629,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 29447
    },
    {
      "epoch": 11.765081901717938,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 29448
    },
    {
      "epoch": 11.765481422293249,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 29449
    },
    {
      "epoch": 11.765880942868558,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 29450
    },
    {
      "epoch": 11.766280463443866,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 29451
    },
    {
      "epoch": 11.766679984019177,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 29452
    },
    {
      "epoch": 11.767079504594486,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 29453
    },
    {
      "epoch": 11.767479025169797,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 29454
    },
    {
      "epoch": 11.767878545745106,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 29455
    },
    {
      "epoch": 11.768278066320416,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 29456
    },
    {
      "epoch": 11.768677586895725,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 29457
    },
    {
      "epoch": 11.769077107471034,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 29458
    },
    {
      "epoch": 11.769476628046345,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 29459
    },
    {
      "epoch": 11.769876148621654,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 29460
    },
    {
      "epoch": 11.770275669196964,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 29461
    },
    {
      "epoch": 11.770675189772273,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 29462
    },
    {
      "epoch": 11.771074710347582,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 29463
    },
    {
      "epoch": 11.771474230922893,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 29464
    },
    {
      "epoch": 11.771873751498202,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 29465
    },
    {
      "epoch": 11.772273272073512,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 29466
    },
    {
      "epoch": 11.772672792648821,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 29467
    },
    {
      "epoch": 11.77307231322413,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 29468
    },
    {
      "epoch": 11.77347183379944,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 29469
    },
    {
      "epoch": 11.77387135437475,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 29470
    },
    {
      "epoch": 11.77427087495006,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 29471
    },
    {
      "epoch": 11.77467039552537,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 29472
    },
    {
      "epoch": 11.77506991610068,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 29473
    },
    {
      "epoch": 11.775469436675989,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 29474
    },
    {
      "epoch": 11.775868957251298,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 29475
    },
    {
      "epoch": 11.776268477826608,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 29476
    },
    {
      "epoch": 11.776667998401917,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 29477
    },
    {
      "epoch": 11.777067518977228,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 29478
    },
    {
      "epoch": 11.777467039552537,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 29479
    },
    {
      "epoch": 11.777866560127846,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 29480
    },
    {
      "epoch": 11.778266080703157,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 29481
    },
    {
      "epoch": 11.778665601278465,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 29482
    },
    {
      "epoch": 11.779065121853776,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 29483
    },
    {
      "epoch": 11.779464642429085,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 29484
    },
    {
      "epoch": 11.779864163004394,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 29485
    },
    {
      "epoch": 11.780263683579705,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 29486
    },
    {
      "epoch": 11.780663204155013,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 29487
    },
    {
      "epoch": 11.781062724730324,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 29488
    },
    {
      "epoch": 11.781462245305633,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 29489
    },
    {
      "epoch": 11.781861765880944,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 29490
    },
    {
      "epoch": 11.782261286456253,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 29491
    },
    {
      "epoch": 11.782660807031561,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 29492
    },
    {
      "epoch": 11.783060327606872,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 29493
    },
    {
      "epoch": 11.783459848182181,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 29494
    },
    {
      "epoch": 11.783859368757492,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 29495
    },
    {
      "epoch": 11.7842588893328,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 29496
    },
    {
      "epoch": 11.78465840990811,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 29497
    },
    {
      "epoch": 11.78505793048342,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 29498
    },
    {
      "epoch": 11.785457451058729,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 29499
    },
    {
      "epoch": 11.78585697163404,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 29500
    },
    {
      "epoch": 11.786256492209349,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 29501
    },
    {
      "epoch": 11.786656012784658,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 29502
    },
    {
      "epoch": 11.787055533359968,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 29503
    },
    {
      "epoch": 11.787455053935277,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 29504
    },
    {
      "epoch": 11.787854574510588,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 29505
    },
    {
      "epoch": 11.788254095085897,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 29506
    },
    {
      "epoch": 11.788653615661207,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 29507
    },
    {
      "epoch": 11.789053136236516,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 29508
    },
    {
      "epoch": 11.789452656811825,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 29509
    },
    {
      "epoch": 11.789852177387136,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 29510
    },
    {
      "epoch": 11.790251697962445,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 29511
    },
    {
      "epoch": 11.790651218537755,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 29512
    },
    {
      "epoch": 11.791050739113064,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 29513
    },
    {
      "epoch": 11.791450259688373,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 29514
    },
    {
      "epoch": 11.791849780263684,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 29515
    },
    {
      "epoch": 11.792249300838993,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 29516
    },
    {
      "epoch": 11.792648821414303,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 29517
    },
    {
      "epoch": 11.793048341989612,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 29518
    },
    {
      "epoch": 11.793447862564921,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 29519
    },
    {
      "epoch": 11.793847383140232,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 29520
    },
    {
      "epoch": 11.79424690371554,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 29521
    },
    {
      "epoch": 11.794646424290852,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 29522
    },
    {
      "epoch": 11.79504594486616,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 29523
    },
    {
      "epoch": 11.795445465441471,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 29524
    },
    {
      "epoch": 11.79584498601678,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 29525
    },
    {
      "epoch": 11.796244506592089,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 29526
    },
    {
      "epoch": 11.7966440271674,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 29527
    },
    {
      "epoch": 11.797043547742708,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 29528
    },
    {
      "epoch": 11.79744306831802,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 29529
    },
    {
      "epoch": 11.797842588893328,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 29530
    },
    {
      "epoch": 11.798242109468637,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 29531
    },
    {
      "epoch": 11.798641630043948,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 29532
    },
    {
      "epoch": 11.799041150619257,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 29533
    },
    {
      "epoch": 11.799440671194567,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 29534
    },
    {
      "epoch": 11.799840191769876,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 29535
    },
    {
      "epoch": 11.800239712345185,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 29536
    },
    {
      "epoch": 11.800639232920496,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 29537
    },
    {
      "epoch": 11.801038753495805,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 29538
    },
    {
      "epoch": 11.801438274071115,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 29539
    },
    {
      "epoch": 11.801837794646424,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 29540
    },
    {
      "epoch": 11.802237315221735,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 29541
    },
    {
      "epoch": 11.802636835797044,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0534,
      "step": 29542
    },
    {
      "epoch": 11.803036356372353,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 29543
    },
    {
      "epoch": 11.803435876947663,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 29544
    },
    {
      "epoch": 11.803835397522972,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 29545
    },
    {
      "epoch": 11.804234918098283,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 29546
    },
    {
      "epoch": 11.804634438673592,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 29547
    },
    {
      "epoch": 11.8050339592489,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 29548
    },
    {
      "epoch": 11.805433479824211,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 29549
    },
    {
      "epoch": 11.80583300039952,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 29550
    },
    {
      "epoch": 11.806232520974831,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 29551
    },
    {
      "epoch": 11.80663204155014,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 29552
    },
    {
      "epoch": 11.807031562125449,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 29553
    },
    {
      "epoch": 11.80743108270076,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 29554
    },
    {
      "epoch": 11.807830603276068,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 29555
    },
    {
      "epoch": 11.808230123851379,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 29556
    },
    {
      "epoch": 11.808629644426688,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 29557
    },
    {
      "epoch": 11.809029165001999,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 29558
    },
    {
      "epoch": 11.809428685577307,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 29559
    },
    {
      "epoch": 11.809828206152616,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 29560
    },
    {
      "epoch": 11.810227726727927,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 29561
    },
    {
      "epoch": 11.810627247303236,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 29562
    },
    {
      "epoch": 11.811026767878547,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 29563
    },
    {
      "epoch": 11.811426288453855,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 29564
    },
    {
      "epoch": 11.811825809029164,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 29565
    },
    {
      "epoch": 11.812225329604475,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 29566
    },
    {
      "epoch": 11.812624850179784,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 29567
    },
    {
      "epoch": 11.813024370755095,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 29568
    },
    {
      "epoch": 11.813423891330403,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 29569
    },
    {
      "epoch": 11.813823411905712,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 29570
    },
    {
      "epoch": 11.814222932481023,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 29571
    },
    {
      "epoch": 11.814622453056332,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 29572
    },
    {
      "epoch": 11.815021973631643,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 29573
    },
    {
      "epoch": 11.815421494206952,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 29574
    },
    {
      "epoch": 11.815821014782262,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 29575
    },
    {
      "epoch": 11.816220535357571,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 29576
    },
    {
      "epoch": 11.81662005593288,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 29577
    },
    {
      "epoch": 11.81701957650819,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 29578
    },
    {
      "epoch": 11.8174190970835,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 29579
    },
    {
      "epoch": 11.81781861765881,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 29580
    },
    {
      "epoch": 11.81821813823412,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 29581
    },
    {
      "epoch": 11.818617658809428,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 29582
    },
    {
      "epoch": 11.819017179384739,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 29583
    },
    {
      "epoch": 11.819416699960048,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 29584
    },
    {
      "epoch": 11.819816220535358,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 29585
    },
    {
      "epoch": 11.820215741110667,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 29586
    },
    {
      "epoch": 11.820615261685976,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 29587
    },
    {
      "epoch": 11.821014782261287,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 29588
    },
    {
      "epoch": 11.821414302836596,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 29589
    },
    {
      "epoch": 11.821813823411906,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 29590
    },
    {
      "epoch": 11.822213343987215,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 29591
    },
    {
      "epoch": 11.822612864562524,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 29592
    },
    {
      "epoch": 11.823012385137835,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 29593
    },
    {
      "epoch": 11.823411905713144,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 29594
    },
    {
      "epoch": 11.823811426288454,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 29595
    },
    {
      "epoch": 11.824210946863763,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 29596
    },
    {
      "epoch": 11.824610467439072,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 29597
    },
    {
      "epoch": 11.825009988014383,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 29598
    },
    {
      "epoch": 11.825409508589692,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 29599
    },
    {
      "epoch": 11.825809029165002,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 29600
    },
    {
      "epoch": 11.826208549740311,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 29601
    },
    {
      "epoch": 11.826608070315622,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 29602
    },
    {
      "epoch": 11.827007590890931,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 29603
    },
    {
      "epoch": 11.82740711146624,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 29604
    },
    {
      "epoch": 11.82780663204155,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 29605
    },
    {
      "epoch": 11.82820615261686,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 29606
    },
    {
      "epoch": 11.82860567319217,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 29607
    },
    {
      "epoch": 11.829005193767479,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 29608
    },
    {
      "epoch": 11.829404714342788,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 29609
    },
    {
      "epoch": 11.829804234918099,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 29610
    },
    {
      "epoch": 11.830203755493407,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 29611
    },
    {
      "epoch": 11.830603276068718,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 29612
    },
    {
      "epoch": 11.831002796644027,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 29613
    },
    {
      "epoch": 11.831402317219336,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 29614
    },
    {
      "epoch": 11.831801837794647,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 29615
    },
    {
      "epoch": 11.832201358369955,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 29616
    },
    {
      "epoch": 11.832600878945266,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 29617
    },
    {
      "epoch": 11.833000399520575,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 29618
    },
    {
      "epoch": 11.833399920095886,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 29619
    },
    {
      "epoch": 11.833799440671195,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 29620
    },
    {
      "epoch": 11.834198961246504,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 29621
    },
    {
      "epoch": 11.834598481821814,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 29622
    },
    {
      "epoch": 11.834998002397123,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 29623
    },
    {
      "epoch": 11.835397522972434,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 29624
    },
    {
      "epoch": 11.835797043547743,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 29625
    },
    {
      "epoch": 11.836196564123052,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 29626
    },
    {
      "epoch": 11.836596084698362,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 29627
    },
    {
      "epoch": 11.836995605273671,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 29628
    },
    {
      "epoch": 11.837395125848982,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 29629
    },
    {
      "epoch": 11.83779464642429,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 29630
    },
    {
      "epoch": 11.8381941669996,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 29631
    },
    {
      "epoch": 11.83859368757491,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 29632
    },
    {
      "epoch": 11.83899320815022,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 29633
    },
    {
      "epoch": 11.83939272872553,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 29634
    },
    {
      "epoch": 11.839792249300839,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 29635
    },
    {
      "epoch": 11.84019176987615,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 29636
    },
    {
      "epoch": 11.840591290451458,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 29637
    },
    {
      "epoch": 11.840990811026767,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 29638
    },
    {
      "epoch": 11.841390331602078,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 29639
    },
    {
      "epoch": 11.841789852177387,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 29640
    },
    {
      "epoch": 11.842189372752697,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 29641
    },
    {
      "epoch": 11.842588893328006,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 29642
    },
    {
      "epoch": 11.842988413903315,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 29643
    },
    {
      "epoch": 11.843387934478626,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 29644
    },
    {
      "epoch": 11.843787455053935,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 29645
    },
    {
      "epoch": 11.844186975629245,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 29646
    },
    {
      "epoch": 11.844586496204554,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 29647
    },
    {
      "epoch": 11.844986016779863,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 29648
    },
    {
      "epoch": 11.845385537355174,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 29649
    },
    {
      "epoch": 11.845785057930483,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 29650
    },
    {
      "epoch": 11.846184578505794,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 29651
    },
    {
      "epoch": 11.846584099081102,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 29652
    },
    {
      "epoch": 11.846983619656413,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 29653
    },
    {
      "epoch": 11.847383140231722,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 29654
    },
    {
      "epoch": 11.847782660807031,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 29655
    },
    {
      "epoch": 11.848182181382342,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 29656
    },
    {
      "epoch": 11.84858170195765,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 29657
    },
    {
      "epoch": 11.848981222532961,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 29658
    },
    {
      "epoch": 11.84938074310827,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 29659
    },
    {
      "epoch": 11.849780263683579,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 29660
    },
    {
      "epoch": 11.85017978425889,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 29661
    },
    {
      "epoch": 11.850579304834199,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 29662
    },
    {
      "epoch": 11.85097882540951,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 29663
    },
    {
      "epoch": 11.851378345984818,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 29664
    },
    {
      "epoch": 11.851777866560127,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 29665
    },
    {
      "epoch": 11.852177387135438,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 29666
    },
    {
      "epoch": 11.852576907710747,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 29667
    },
    {
      "epoch": 11.852976428286057,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 29668
    },
    {
      "epoch": 11.853375948861366,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 29669
    },
    {
      "epoch": 11.853775469436677,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 29670
    },
    {
      "epoch": 11.854174990011986,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 29671
    },
    {
      "epoch": 11.854574510587295,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 29672
    },
    {
      "epoch": 11.854974031162605,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 29673
    },
    {
      "epoch": 11.855373551737914,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 29674
    },
    {
      "epoch": 11.855773072313225,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 29675
    },
    {
      "epoch": 11.856172592888534,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 29676
    },
    {
      "epoch": 11.856572113463843,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 29677
    },
    {
      "epoch": 11.856971634039153,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 29678
    },
    {
      "epoch": 11.857371154614462,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 29679
    },
    {
      "epoch": 11.857770675189773,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 29680
    },
    {
      "epoch": 11.858170195765082,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 29681
    },
    {
      "epoch": 11.85856971634039,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 29682
    },
    {
      "epoch": 11.858969236915701,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 29683
    },
    {
      "epoch": 11.85936875749101,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 29684
    },
    {
      "epoch": 11.859768278066321,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 29685
    },
    {
      "epoch": 11.86016779864163,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 29686
    },
    {
      "epoch": 11.86056731921694,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 29687
    },
    {
      "epoch": 11.86096683979225,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 29688
    },
    {
      "epoch": 11.861366360367558,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 29689
    },
    {
      "epoch": 11.861765880942869,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 29690
    },
    {
      "epoch": 11.862165401518178,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 29691
    },
    {
      "epoch": 11.862564922093489,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 29692
    },
    {
      "epoch": 11.862964442668797,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 29693
    },
    {
      "epoch": 11.863363963244106,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 29694
    },
    {
      "epoch": 11.863763483819417,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 29695
    },
    {
      "epoch": 11.864163004394726,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 29696
    },
    {
      "epoch": 11.864562524970037,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 29697
    },
    {
      "epoch": 11.864962045545346,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 29698
    },
    {
      "epoch": 11.865361566120654,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 29699
    },
    {
      "epoch": 11.865761086695965,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 29700
    },
    {
      "epoch": 11.866160607271274,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 29701
    },
    {
      "epoch": 11.866560127846585,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0582,
      "step": 29702
    },
    {
      "epoch": 11.866959648421894,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 29703
    },
    {
      "epoch": 11.867359168997204,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 29704
    },
    {
      "epoch": 11.867758689572513,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 29705
    },
    {
      "epoch": 11.868158210147822,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 29706
    },
    {
      "epoch": 11.868557730723133,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 29707
    },
    {
      "epoch": 11.868957251298442,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 29708
    },
    {
      "epoch": 11.869356771873752,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 29709
    },
    {
      "epoch": 11.869756292449061,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 29710
    },
    {
      "epoch": 11.87015581302437,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 29711
    },
    {
      "epoch": 11.87055533359968,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 29712
    },
    {
      "epoch": 11.87095485417499,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 29713
    },
    {
      "epoch": 11.8713543747503,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 29714
    },
    {
      "epoch": 11.87175389532561,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 29715
    },
    {
      "epoch": 11.872153415900918,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 29716
    },
    {
      "epoch": 11.872552936476229,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 29717
    },
    {
      "epoch": 11.872952457051538,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 29718
    },
    {
      "epoch": 11.873351977626848,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 29719
    },
    {
      "epoch": 11.873751498202157,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 29720
    },
    {
      "epoch": 11.874151018777468,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 29721
    },
    {
      "epoch": 11.874550539352777,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 29722
    },
    {
      "epoch": 11.874950059928086,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 29723
    },
    {
      "epoch": 11.875349580503396,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 29724
    },
    {
      "epoch": 11.875749101078705,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 29725
    },
    {
      "epoch": 11.876148621654016,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 29726
    },
    {
      "epoch": 11.876548142229325,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 29727
    },
    {
      "epoch": 11.876947662804634,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 29728
    },
    {
      "epoch": 11.877347183379944,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 29729
    },
    {
      "epoch": 11.877746703955253,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 29730
    },
    {
      "epoch": 11.878146224530564,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 29731
    },
    {
      "epoch": 11.878545745105873,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 29732
    },
    {
      "epoch": 11.878945265681182,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 29733
    },
    {
      "epoch": 11.879344786256492,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 29734
    },
    {
      "epoch": 11.879744306831801,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 29735
    },
    {
      "epoch": 11.880143827407112,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 29736
    },
    {
      "epoch": 11.880543347982421,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 29737
    },
    {
      "epoch": 11.880942868557732,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 29738
    },
    {
      "epoch": 11.88134238913304,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 29739
    },
    {
      "epoch": 11.88174190970835,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 29740
    },
    {
      "epoch": 11.88214143028366,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 29741
    },
    {
      "epoch": 11.882540950858969,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 29742
    },
    {
      "epoch": 11.88294047143428,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 29743
    },
    {
      "epoch": 11.883339992009589,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 29744
    },
    {
      "epoch": 11.883739512584897,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 29745
    },
    {
      "epoch": 11.884139033160208,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 29746
    },
    {
      "epoch": 11.884538553735517,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 29747
    },
    {
      "epoch": 11.884938074310828,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 29748
    },
    {
      "epoch": 11.885337594886137,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 29749
    },
    {
      "epoch": 11.885737115461446,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 29750
    },
    {
      "epoch": 11.886136636036756,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 29751
    },
    {
      "epoch": 11.886536156612065,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 29752
    },
    {
      "epoch": 11.886935677187376,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 29753
    },
    {
      "epoch": 11.887335197762685,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 29754
    },
    {
      "epoch": 11.887734718337995,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 29755
    },
    {
      "epoch": 11.888134238913304,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 29756
    },
    {
      "epoch": 11.888533759488613,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 29757
    },
    {
      "epoch": 11.888933280063924,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 29758
    },
    {
      "epoch": 11.889332800639233,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 29759
    },
    {
      "epoch": 11.889732321214543,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 29760
    },
    {
      "epoch": 11.890131841789852,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 29761
    },
    {
      "epoch": 11.890531362365161,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 29762
    },
    {
      "epoch": 11.890930882940472,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 29763
    },
    {
      "epoch": 11.89133040351578,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 29764
    },
    {
      "epoch": 11.891729924091091,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 29765
    },
    {
      "epoch": 11.8921294446664,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 29766
    },
    {
      "epoch": 11.89252896524171,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 29767
    },
    {
      "epoch": 11.89292848581702,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 29768
    },
    {
      "epoch": 11.893328006392329,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 29769
    },
    {
      "epoch": 11.89372752696764,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 29770
    },
    {
      "epoch": 11.894127047542948,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 29771
    },
    {
      "epoch": 11.894526568118257,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 29772
    },
    {
      "epoch": 11.894926088693568,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 29773
    },
    {
      "epoch": 11.895325609268877,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 29774
    },
    {
      "epoch": 11.895725129844188,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 29775
    },
    {
      "epoch": 11.896124650419496,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 29776
    },
    {
      "epoch": 11.896524170994805,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 29777
    },
    {
      "epoch": 11.896923691570116,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 29778
    },
    {
      "epoch": 11.897323212145425,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 29779
    },
    {
      "epoch": 11.897722732720736,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 29780
    },
    {
      "epoch": 11.898122253296044,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 29781
    },
    {
      "epoch": 11.898521773871355,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 29782
    },
    {
      "epoch": 11.898921294446664,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 29783
    },
    {
      "epoch": 11.899320815021973,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 29784
    },
    {
      "epoch": 11.899720335597284,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 29785
    },
    {
      "epoch": 11.900119856172592,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 29786
    },
    {
      "epoch": 11.900519376747903,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 29787
    },
    {
      "epoch": 11.900918897323212,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 29788
    },
    {
      "epoch": 11.901318417898521,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 29789
    },
    {
      "epoch": 11.901717938473832,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 29790
    },
    {
      "epoch": 11.90211745904914,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 29791
    },
    {
      "epoch": 11.902516979624451,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 29792
    },
    {
      "epoch": 11.90291650019976,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 29793
    },
    {
      "epoch": 11.903316020775069,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 29794
    },
    {
      "epoch": 11.90371554135038,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 29795
    },
    {
      "epoch": 11.904115061925689,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 29796
    },
    {
      "epoch": 11.904514582501,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 29797
    },
    {
      "epoch": 11.904914103076308,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 29798
    },
    {
      "epoch": 11.905313623651619,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 29799
    },
    {
      "epoch": 11.905713144226928,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 29800
    },
    {
      "epoch": 11.906112664802237,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 29801
    },
    {
      "epoch": 11.906512185377547,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 29802
    },
    {
      "epoch": 11.906911705952856,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 29803
    },
    {
      "epoch": 11.907311226528167,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 29804
    },
    {
      "epoch": 11.907710747103476,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 29805
    },
    {
      "epoch": 11.908110267678785,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 29806
    },
    {
      "epoch": 11.908509788254095,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 29807
    },
    {
      "epoch": 11.908909308829404,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 29808
    },
    {
      "epoch": 11.909308829404715,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 29809
    },
    {
      "epoch": 11.909708349980024,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 29810
    },
    {
      "epoch": 11.910107870555333,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 29811
    },
    {
      "epoch": 11.910507391130643,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 29812
    },
    {
      "epoch": 11.910906911705952,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 29813
    },
    {
      "epoch": 11.911306432281263,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 29814
    },
    {
      "epoch": 11.911705952856572,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 29815
    },
    {
      "epoch": 11.912105473431883,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 29816
    },
    {
      "epoch": 11.912504994007191,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 29817
    },
    {
      "epoch": 11.9129045145825,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 29818
    },
    {
      "epoch": 11.913304035157811,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 29819
    },
    {
      "epoch": 11.91370355573312,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 29820
    },
    {
      "epoch": 11.91410307630843,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 29821
    },
    {
      "epoch": 11.91450259688374,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 29822
    },
    {
      "epoch": 11.914902117459048,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 29823
    },
    {
      "epoch": 11.915301638034359,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 29824
    },
    {
      "epoch": 11.915701158609668,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 29825
    },
    {
      "epoch": 11.916100679184979,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 29826
    },
    {
      "epoch": 11.916500199760288,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 29827
    },
    {
      "epoch": 11.916899720335596,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 29828
    },
    {
      "epoch": 11.917299240910907,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 29829
    },
    {
      "epoch": 11.917698761486216,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 29830
    },
    {
      "epoch": 11.918098282061527,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 29831
    },
    {
      "epoch": 11.918497802636836,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 29832
    },
    {
      "epoch": 11.918897323212146,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 29833
    },
    {
      "epoch": 11.919296843787455,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 29834
    },
    {
      "epoch": 11.919696364362764,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 29835
    },
    {
      "epoch": 11.920095884938075,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 29836
    },
    {
      "epoch": 11.920495405513384,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 29837
    },
    {
      "epoch": 11.920894926088694,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 29838
    },
    {
      "epoch": 11.921294446664003,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 29839
    },
    {
      "epoch": 11.921693967239312,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 29840
    },
    {
      "epoch": 11.922093487814623,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 29841
    },
    {
      "epoch": 11.922493008389932,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 29842
    },
    {
      "epoch": 11.922892528965242,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 29843
    },
    {
      "epoch": 11.923292049540551,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 29844
    },
    {
      "epoch": 11.92369157011586,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 29845
    },
    {
      "epoch": 11.92409109069117,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 29846
    },
    {
      "epoch": 11.92449061126648,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 29847
    },
    {
      "epoch": 11.92489013184179,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 29848
    },
    {
      "epoch": 11.9252896524171,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 29849
    },
    {
      "epoch": 11.92568917299241,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 29850
    },
    {
      "epoch": 11.926088693567719,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 29851
    },
    {
      "epoch": 11.926488214143028,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 29852
    },
    {
      "epoch": 11.926887734718338,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 29853
    },
    {
      "epoch": 11.927287255293647,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 29854
    },
    {
      "epoch": 11.927686775868958,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 29855
    },
    {
      "epoch": 11.928086296444267,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 29856
    },
    {
      "epoch": 11.928485817019576,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 29857
    },
    {
      "epoch": 11.928885337594886,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 29858
    },
    {
      "epoch": 11.929284858170195,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 29859
    },
    {
      "epoch": 11.929684378745506,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 29860
    },
    {
      "epoch": 11.930083899320815,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 29861
    },
    {
      "epoch": 11.930483419896124,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 29862
    },
    {
      "epoch": 11.930882940471434,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 29863
    },
    {
      "epoch": 11.931282461046743,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 29864
    },
    {
      "epoch": 11.931681981622054,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 29865
    },
    {
      "epoch": 11.932081502197363,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 29866
    },
    {
      "epoch": 11.932481022772674,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 29867
    },
    {
      "epoch": 11.932880543347983,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 29868
    },
    {
      "epoch": 11.933280063923291,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 29869
    },
    {
      "epoch": 11.933679584498602,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 29870
    },
    {
      "epoch": 11.934079105073911,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 29871
    },
    {
      "epoch": 11.934478625649222,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 29872
    },
    {
      "epoch": 11.93487814622453,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 29873
    },
    {
      "epoch": 11.93527766679984,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 29874
    },
    {
      "epoch": 11.93567718737515,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 29875
    },
    {
      "epoch": 11.936076707950459,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 29876
    },
    {
      "epoch": 11.93647622852577,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 29877
    },
    {
      "epoch": 11.936875749101079,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 29878
    },
    {
      "epoch": 11.937275269676388,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 29879
    },
    {
      "epoch": 11.937674790251698,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 29880
    },
    {
      "epoch": 11.938074310827007,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 29881
    },
    {
      "epoch": 11.938473831402318,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 29882
    },
    {
      "epoch": 11.938873351977627,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 29883
    },
    {
      "epoch": 11.939272872552937,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 29884
    },
    {
      "epoch": 11.939672393128246,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 29885
    },
    {
      "epoch": 11.940071913703555,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 29886
    },
    {
      "epoch": 11.940471434278866,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 29887
    },
    {
      "epoch": 11.940870954854175,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 29888
    },
    {
      "epoch": 11.941270475429485,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 29889
    },
    {
      "epoch": 11.941669996004794,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 29890
    },
    {
      "epoch": 11.942069516580103,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 29891
    },
    {
      "epoch": 11.942469037155414,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 29892
    },
    {
      "epoch": 11.942868557730723,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 29893
    },
    {
      "epoch": 11.943268078306033,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 29894
    },
    {
      "epoch": 11.943667598881342,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 29895
    },
    {
      "epoch": 11.944067119456651,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 29896
    },
    {
      "epoch": 11.944466640031962,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 29897
    },
    {
      "epoch": 11.94486616060727,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 29898
    },
    {
      "epoch": 11.945265681182581,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 29899
    },
    {
      "epoch": 11.94566520175789,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 29900
    },
    {
      "epoch": 11.946064722333201,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 29901
    },
    {
      "epoch": 11.94646424290851,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 29902
    },
    {
      "epoch": 11.946863763483819,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 29903
    },
    {
      "epoch": 11.94726328405913,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 29904
    },
    {
      "epoch": 11.947662804634438,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 29905
    },
    {
      "epoch": 11.948062325209749,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 29906
    },
    {
      "epoch": 11.948461845785058,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 29907
    },
    {
      "epoch": 11.948861366360367,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 29908
    },
    {
      "epoch": 11.949260886935678,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 29909
    },
    {
      "epoch": 11.949660407510986,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 29910
    },
    {
      "epoch": 11.950059928086297,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 29911
    },
    {
      "epoch": 11.950459448661606,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 29912
    },
    {
      "epoch": 11.950858969236915,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 29913
    },
    {
      "epoch": 11.951258489812226,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 29914
    },
    {
      "epoch": 11.951658010387535,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 29915
    },
    {
      "epoch": 11.952057530962845,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 29916
    },
    {
      "epoch": 11.952457051538154,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 29917
    },
    {
      "epoch": 11.952856572113465,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 29918
    },
    {
      "epoch": 11.953256092688774,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 29919
    },
    {
      "epoch": 11.953655613264083,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 29920
    },
    {
      "epoch": 11.954055133839393,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 29921
    },
    {
      "epoch": 11.954454654414702,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 29922
    },
    {
      "epoch": 11.954854174990013,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 29923
    },
    {
      "epoch": 11.955253695565322,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 29924
    },
    {
      "epoch": 11.95565321614063,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 29925
    },
    {
      "epoch": 11.956052736715941,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 29926
    },
    {
      "epoch": 11.95645225729125,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 29927
    },
    {
      "epoch": 11.95685177786656,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 29928
    },
    {
      "epoch": 11.95725129844187,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 29929
    },
    {
      "epoch": 11.957650819017179,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 29930
    },
    {
      "epoch": 11.95805033959249,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 29931
    },
    {
      "epoch": 11.958449860167798,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 29932
    },
    {
      "epoch": 11.958849380743109,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 29933
    },
    {
      "epoch": 11.959248901318418,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 29934
    },
    {
      "epoch": 11.959648421893728,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 29935
    },
    {
      "epoch": 11.960047942469037,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 29936
    },
    {
      "epoch": 11.960447463044346,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 29937
    },
    {
      "epoch": 11.960846983619657,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 29938
    },
    {
      "epoch": 11.961246504194966,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 29939
    },
    {
      "epoch": 11.961646024770276,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 29940
    },
    {
      "epoch": 11.962045545345585,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 29941
    },
    {
      "epoch": 11.962445065920894,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 29942
    },
    {
      "epoch": 11.962844586496205,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 29943
    },
    {
      "epoch": 11.963244107071514,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 29944
    },
    {
      "epoch": 11.963643627646825,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 29945
    },
    {
      "epoch": 11.964043148222133,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 29946
    },
    {
      "epoch": 11.964442668797442,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 29947
    },
    {
      "epoch": 11.964842189372753,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 29948
    },
    {
      "epoch": 11.965241709948062,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 29949
    },
    {
      "epoch": 11.965641230523373,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 29950
    },
    {
      "epoch": 11.966040751098681,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 29951
    },
    {
      "epoch": 11.96644027167399,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 29952
    },
    {
      "epoch": 11.966839792249301,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 29953
    },
    {
      "epoch": 11.96723931282461,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0108,
      "step": 29954
    },
    {
      "epoch": 11.96763883339992,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 29955
    },
    {
      "epoch": 11.96803835397523,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 29956
    },
    {
      "epoch": 11.968437874550538,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 29957
    },
    {
      "epoch": 11.968837395125849,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 29958
    },
    {
      "epoch": 11.969236915701158,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 29959
    },
    {
      "epoch": 11.969636436276469,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 29960
    },
    {
      "epoch": 11.970035956851778,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 29961
    },
    {
      "epoch": 11.970435477427088,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 29962
    },
    {
      "epoch": 11.970834998002397,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 29963
    },
    {
      "epoch": 11.971234518577706,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 29964
    },
    {
      "epoch": 11.971634039153017,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 29965
    },
    {
      "epoch": 11.972033559728326,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 29966
    },
    {
      "epoch": 11.972433080303636,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 29967
    },
    {
      "epoch": 11.972832600878945,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 29968
    },
    {
      "epoch": 11.973232121454254,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 29969
    },
    {
      "epoch": 11.973631642029565,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 29970
    },
    {
      "epoch": 11.974031162604874,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 29971
    },
    {
      "epoch": 11.974430683180184,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 29972
    },
    {
      "epoch": 11.974830203755493,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 29973
    },
    {
      "epoch": 11.975229724330802,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 29974
    },
    {
      "epoch": 11.975629244906113,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 29975
    },
    {
      "epoch": 11.976028765481422,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 29976
    },
    {
      "epoch": 11.976428286056732,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 29977
    },
    {
      "epoch": 11.976827806632041,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 29978
    },
    {
      "epoch": 11.977227327207352,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 29979
    },
    {
      "epoch": 11.97762684778266,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 29980
    },
    {
      "epoch": 11.97802636835797,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 29981
    },
    {
      "epoch": 11.97842588893328,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 29982
    },
    {
      "epoch": 11.97882540950859,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 29983
    },
    {
      "epoch": 11.9792249300839,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 29984
    },
    {
      "epoch": 11.979624450659209,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 29985
    },
    {
      "epoch": 11.980023971234518,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 29986
    },
    {
      "epoch": 11.980423491809828,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 29987
    },
    {
      "epoch": 11.980823012385137,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 29988
    },
    {
      "epoch": 11.981222532960448,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 29989
    },
    {
      "epoch": 11.981622053535757,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 29990
    },
    {
      "epoch": 11.982021574111066,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 29991
    },
    {
      "epoch": 11.982421094686377,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 29992
    },
    {
      "epoch": 11.982820615261685,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 29993
    },
    {
      "epoch": 11.983220135836996,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 29994
    },
    {
      "epoch": 11.983619656412305,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 29995
    },
    {
      "epoch": 11.984019176987616,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 29996
    },
    {
      "epoch": 11.984418697562925,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 29997
    },
    {
      "epoch": 11.984818218138233,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 29998
    },
    {
      "epoch": 11.985217738713544,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 29999
    },
    {
      "epoch": 11.985617259288853,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 30000
    },
    {
      "epoch": 11.986016779864164,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 30001
    },
    {
      "epoch": 11.986416300439473,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 30002
    },
    {
      "epoch": 11.986815821014781,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 30003
    },
    {
      "epoch": 11.987215341590092,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 30004
    },
    {
      "epoch": 11.987614862165401,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 30005
    },
    {
      "epoch": 11.988014382740712,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 30006
    },
    {
      "epoch": 11.98841390331602,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 30007
    },
    {
      "epoch": 11.98881342389133,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 30008
    },
    {
      "epoch": 11.98921294446664,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 30009
    },
    {
      "epoch": 11.989612465041949,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 30010
    },
    {
      "epoch": 11.99001198561726,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 30011
    },
    {
      "epoch": 11.990411506192569,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 30012
    },
    {
      "epoch": 11.99081102676788,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 30013
    },
    {
      "epoch": 11.991210547343188,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 30014
    },
    {
      "epoch": 11.991610067918497,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 30015
    },
    {
      "epoch": 11.992009588493808,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 30016
    },
    {
      "epoch": 11.992409109069117,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 30017
    },
    {
      "epoch": 11.992808629644427,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 30018
    },
    {
      "epoch": 11.993208150219736,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 30019
    },
    {
      "epoch": 11.993607670795045,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 30020
    },
    {
      "epoch": 11.994007191370356,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 30021
    },
    {
      "epoch": 11.994406711945665,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 30022
    },
    {
      "epoch": 11.994806232520975,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 30023
    },
    {
      "epoch": 11.995205753096284,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 30024
    },
    {
      "epoch": 11.995605273671593,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 30025
    },
    {
      "epoch": 11.996004794246904,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 30026
    },
    {
      "epoch": 11.996404314822213,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 30027
    },
    {
      "epoch": 11.996803835397523,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 30028
    },
    {
      "epoch": 11.997203355972832,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 30029
    },
    {
      "epoch": 11.997602876548143,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 30030
    },
    {
      "epoch": 11.998002397123452,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 30031
    },
    {
      "epoch": 11.99840191769876,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 30032
    },
    {
      "epoch": 11.998801438274072,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 30033
    },
    {
      "epoch": 11.99920095884938,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 30034
    },
    {
      "epoch": 11.999600479424691,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 30035
    },
    {
      "epoch": 12.0,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 30036
    },
    {
      "epoch": 12.000399520575309,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 30037
    },
    {
      "epoch": 12.00079904115062,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 30038
    },
    {
      "epoch": 12.001198561725928,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 30039
    },
    {
      "epoch": 12.00159808230124,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 30040
    },
    {
      "epoch": 12.001997602876548,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 30041
    },
    {
      "epoch": 12.002397123451857,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 30042
    },
    {
      "epoch": 12.002796644027168,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 30043
    },
    {
      "epoch": 12.003196164602477,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 30044
    },
    {
      "epoch": 12.003595685177787,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 30045
    },
    {
      "epoch": 12.003995205753096,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 30046
    },
    {
      "epoch": 12.004394726328407,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 30047
    },
    {
      "epoch": 12.004794246903716,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 30048
    },
    {
      "epoch": 12.005193767479025,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 30049
    },
    {
      "epoch": 12.005593288054335,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 30050
    },
    {
      "epoch": 12.005992808629644,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 30051
    },
    {
      "epoch": 12.006392329204955,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 30052
    },
    {
      "epoch": 12.006791849780264,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 30053
    },
    {
      "epoch": 12.007191370355573,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 30054
    },
    {
      "epoch": 12.007590890930883,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 30055
    },
    {
      "epoch": 12.007990411506192,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 30056
    },
    {
      "epoch": 12.008389932081503,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 30057
    },
    {
      "epoch": 12.008789452656812,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 30058
    },
    {
      "epoch": 12.00918897323212,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 30059
    },
    {
      "epoch": 12.009588493807431,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 30060
    },
    {
      "epoch": 12.00998801438274,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 30061
    },
    {
      "epoch": 12.010387534958051,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 30062
    },
    {
      "epoch": 12.01078705553336,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 30063
    },
    {
      "epoch": 12.01118657610867,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 30064
    },
    {
      "epoch": 12.01158609668398,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 30065
    },
    {
      "epoch": 12.011985617259288,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 30066
    },
    {
      "epoch": 12.012385137834599,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 30067
    },
    {
      "epoch": 12.012784658409908,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 30068
    },
    {
      "epoch": 12.013184178985219,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 30069
    },
    {
      "epoch": 12.013583699560527,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 30070
    },
    {
      "epoch": 12.013983220135836,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 30071
    },
    {
      "epoch": 12.014382740711147,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 30072
    },
    {
      "epoch": 12.014782261286456,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 30073
    },
    {
      "epoch": 12.015181781861767,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 30074
    },
    {
      "epoch": 12.015581302437075,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 30075
    },
    {
      "epoch": 12.015980823012384,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 30076
    },
    {
      "epoch": 12.016380343587695,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 30077
    },
    {
      "epoch": 12.016779864163004,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 30078
    },
    {
      "epoch": 12.017179384738315,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 30079
    },
    {
      "epoch": 12.017578905313623,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 30080
    },
    {
      "epoch": 12.017978425888934,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 30081
    },
    {
      "epoch": 12.018377946464243,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 30082
    },
    {
      "epoch": 12.018777467039552,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0509,
      "step": 30083
    },
    {
      "epoch": 12.019176987614863,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 30084
    },
    {
      "epoch": 12.019576508190172,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 30085
    },
    {
      "epoch": 12.019976028765482,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 30086
    },
    {
      "epoch": 12.020375549340791,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 30087
    },
    {
      "epoch": 12.0207750699161,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 30088
    },
    {
      "epoch": 12.02117459049141,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 30089
    },
    {
      "epoch": 12.02157411106672,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 30090
    },
    {
      "epoch": 12.02197363164203,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 30091
    },
    {
      "epoch": 12.02237315221734,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 30092
    },
    {
      "epoch": 12.022772672792648,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 30093
    },
    {
      "epoch": 12.023172193367959,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 30094
    },
    {
      "epoch": 12.023571713943268,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 30095
    },
    {
      "epoch": 12.023971234518578,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 30096
    },
    {
      "epoch": 12.024370755093887,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 30097
    },
    {
      "epoch": 12.024770275669196,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 30098
    },
    {
      "epoch": 12.025169796244507,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 30099
    },
    {
      "epoch": 12.025569316819816,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 30100
    },
    {
      "epoch": 12.025968837395126,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 30101
    },
    {
      "epoch": 12.026368357970435,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 30102
    },
    {
      "epoch": 12.026767878545746,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 30103
    },
    {
      "epoch": 12.027167399121055,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 30104
    },
    {
      "epoch": 12.027566919696364,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 30105
    },
    {
      "epoch": 12.027966440271674,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 30106
    },
    {
      "epoch": 12.028365960846983,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 30107
    },
    {
      "epoch": 12.028765481422294,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 30108
    },
    {
      "epoch": 12.029165001997603,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 30109
    },
    {
      "epoch": 12.029564522572912,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 30110
    },
    {
      "epoch": 12.029964043148222,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 30111
    },
    {
      "epoch": 12.030363563723531,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 30112
    },
    {
      "epoch": 12.030763084298842,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 30113
    },
    {
      "epoch": 12.031162604874151,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 30114
    },
    {
      "epoch": 12.03156212544946,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 30115
    },
    {
      "epoch": 12.03196164602477,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 30116
    },
    {
      "epoch": 12.03236116660008,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 30117
    },
    {
      "epoch": 12.03276068717539,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 30118
    },
    {
      "epoch": 12.033160207750699,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 30119
    },
    {
      "epoch": 12.03355972832601,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 30120
    },
    {
      "epoch": 12.033959248901319,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 30121
    },
    {
      "epoch": 12.034358769476627,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 30122
    },
    {
      "epoch": 12.034758290051938,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 30123
    },
    {
      "epoch": 12.035157810627247,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 30124
    },
    {
      "epoch": 12.035557331202558,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 30125
    },
    {
      "epoch": 12.035956851777867,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 30126
    },
    {
      "epoch": 12.036356372353175,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 30127
    },
    {
      "epoch": 12.036755892928486,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 30128
    },
    {
      "epoch": 12.037155413503795,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 30129
    },
    {
      "epoch": 12.037554934079106,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 30130
    },
    {
      "epoch": 12.037954454654415,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 30131
    },
    {
      "epoch": 12.038353975229724,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 30132
    },
    {
      "epoch": 12.038753495805034,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 30133
    },
    {
      "epoch": 12.039153016380343,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 30134
    },
    {
      "epoch": 12.039552536955654,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 30135
    },
    {
      "epoch": 12.039952057530963,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 30136
    },
    {
      "epoch": 12.040351578106273,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 30137
    },
    {
      "epoch": 12.040751098681582,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 30138
    },
    {
      "epoch": 12.041150619256891,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 30139
    },
    {
      "epoch": 12.041550139832202,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 30140
    },
    {
      "epoch": 12.04194966040751,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 30141
    },
    {
      "epoch": 12.042349180982821,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 30142
    },
    {
      "epoch": 12.04274870155813,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 30143
    },
    {
      "epoch": 12.04314822213344,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 30144
    },
    {
      "epoch": 12.04354774270875,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 30145
    },
    {
      "epoch": 12.043947263284059,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 30146
    },
    {
      "epoch": 12.04434678385937,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 30147
    },
    {
      "epoch": 12.044746304434678,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 30148
    },
    {
      "epoch": 12.045145825009987,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 30149
    },
    {
      "epoch": 12.045545345585298,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 30150
    },
    {
      "epoch": 12.045944866160607,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 30151
    },
    {
      "epoch": 12.046344386735917,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 30152
    },
    {
      "epoch": 12.046743907311226,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 30153
    },
    {
      "epoch": 12.047143427886537,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 30154
    },
    {
      "epoch": 12.047542948461846,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 30155
    },
    {
      "epoch": 12.047942469037155,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 30156
    },
    {
      "epoch": 12.048341989612465,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 30157
    },
    {
      "epoch": 12.048741510187774,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 30158
    },
    {
      "epoch": 12.049141030763085,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 30159
    },
    {
      "epoch": 12.049540551338394,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 30160
    },
    {
      "epoch": 12.049940071913703,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 30161
    },
    {
      "epoch": 12.050339592489014,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 30162
    },
    {
      "epoch": 12.050739113064322,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 30163
    },
    {
      "epoch": 12.051138633639633,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 30164
    },
    {
      "epoch": 12.051538154214942,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 30165
    },
    {
      "epoch": 12.051937674790251,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 30166
    },
    {
      "epoch": 12.052337195365562,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 30167
    },
    {
      "epoch": 12.05273671594087,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 30168
    },
    {
      "epoch": 12.053136236516181,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 30169
    },
    {
      "epoch": 12.05353575709149,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 30170
    },
    {
      "epoch": 12.053935277666799,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 30171
    },
    {
      "epoch": 12.05433479824211,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 30172
    },
    {
      "epoch": 12.054734318817419,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 30173
    },
    {
      "epoch": 12.05513383939273,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 30174
    },
    {
      "epoch": 12.055533359968038,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 30175
    },
    {
      "epoch": 12.055932880543349,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 30176
    },
    {
      "epoch": 12.056332401118658,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 30177
    },
    {
      "epoch": 12.056731921693967,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 30178
    },
    {
      "epoch": 12.057131442269277,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 30179
    },
    {
      "epoch": 12.057530962844586,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 30180
    },
    {
      "epoch": 12.057930483419897,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 30181
    },
    {
      "epoch": 12.058330003995206,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 30182
    },
    {
      "epoch": 12.058729524570515,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 30183
    },
    {
      "epoch": 12.059129045145825,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0479,
      "step": 30184
    },
    {
      "epoch": 12.059528565721134,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 30185
    },
    {
      "epoch": 12.059928086296445,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 30186
    },
    {
      "epoch": 12.060327606871754,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 30187
    },
    {
      "epoch": 12.060727127447063,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 30188
    },
    {
      "epoch": 12.061126648022373,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 30189
    },
    {
      "epoch": 12.061526168597682,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 30190
    },
    {
      "epoch": 12.061925689172993,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 30191
    },
    {
      "epoch": 12.062325209748302,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 30192
    },
    {
      "epoch": 12.062724730323612,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 30193
    },
    {
      "epoch": 12.063124250898921,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 30194
    },
    {
      "epoch": 12.06352377147423,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 30195
    },
    {
      "epoch": 12.063923292049541,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 30196
    },
    {
      "epoch": 12.06432281262485,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 30197
    },
    {
      "epoch": 12.06472233320016,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 30198
    },
    {
      "epoch": 12.06512185377547,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 30199
    },
    {
      "epoch": 12.065521374350778,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 30200
    },
    {
      "epoch": 12.065920894926089,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 30201
    },
    {
      "epoch": 12.066320415501398,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 30202
    },
    {
      "epoch": 12.066719936076709,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 30203
    },
    {
      "epoch": 12.067119456652017,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 30204
    },
    {
      "epoch": 12.067518977227326,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 30205
    },
    {
      "epoch": 12.067918497802637,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 30206
    },
    {
      "epoch": 12.068318018377946,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 30207
    },
    {
      "epoch": 12.068717538953257,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 30208
    },
    {
      "epoch": 12.069117059528566,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 30209
    },
    {
      "epoch": 12.069516580103876,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 30210
    },
    {
      "epoch": 12.069916100679185,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 30211
    },
    {
      "epoch": 12.070315621254494,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 30212
    },
    {
      "epoch": 12.070715141829805,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 30213
    },
    {
      "epoch": 12.071114662405114,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 30214
    },
    {
      "epoch": 12.071514182980424,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 30215
    },
    {
      "epoch": 12.071913703555733,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 30216
    },
    {
      "epoch": 12.072313224131042,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 30217
    },
    {
      "epoch": 12.072712744706353,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 30218
    },
    {
      "epoch": 12.073112265281662,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 30219
    },
    {
      "epoch": 12.073511785856972,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 30220
    },
    {
      "epoch": 12.073911306432281,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 30221
    },
    {
      "epoch": 12.07431082700759,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 30222
    },
    {
      "epoch": 12.0747103475829,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 30223
    },
    {
      "epoch": 12.07510986815821,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 30224
    },
    {
      "epoch": 12.07550938873352,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 30225
    },
    {
      "epoch": 12.07590890930883,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 30226
    },
    {
      "epoch": 12.07630842988414,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 30227
    },
    {
      "epoch": 12.076707950459449,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 30228
    },
    {
      "epoch": 12.077107471034758,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 30229
    },
    {
      "epoch": 12.077506991610068,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 30230
    },
    {
      "epoch": 12.077906512185377,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 30231
    },
    {
      "epoch": 12.078306032760688,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 30232
    },
    {
      "epoch": 12.078705553335997,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 30233
    },
    {
      "epoch": 12.079105073911306,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 30234
    },
    {
      "epoch": 12.079504594486616,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 30235
    },
    {
      "epoch": 12.079904115061925,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 30236
    },
    {
      "epoch": 12.080303635637236,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 30237
    },
    {
      "epoch": 12.080703156212545,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 30238
    },
    {
      "epoch": 12.081102676787854,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 30239
    },
    {
      "epoch": 12.081502197363164,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 30240
    },
    {
      "epoch": 12.081901717938473,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 30241
    },
    {
      "epoch": 12.082301238513784,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 30242
    },
    {
      "epoch": 12.082700759089093,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 30243
    },
    {
      "epoch": 12.083100279664404,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 30244
    },
    {
      "epoch": 12.083499800239712,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 30245
    },
    {
      "epoch": 12.083899320815021,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 30246
    },
    {
      "epoch": 12.084298841390332,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 30247
    },
    {
      "epoch": 12.084698361965641,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 30248
    },
    {
      "epoch": 12.085097882540952,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 30249
    },
    {
      "epoch": 12.08549740311626,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 30250
    },
    {
      "epoch": 12.08589692369157,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 30251
    },
    {
      "epoch": 12.08629644426688,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 30252
    },
    {
      "epoch": 12.086695964842189,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 30253
    },
    {
      "epoch": 12.0870954854175,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 30254
    },
    {
      "epoch": 12.087495005992809,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 30255
    },
    {
      "epoch": 12.087894526568117,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 30256
    },
    {
      "epoch": 12.088294047143428,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 30257
    },
    {
      "epoch": 12.088693567718737,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 30258
    },
    {
      "epoch": 12.089093088294048,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 30259
    },
    {
      "epoch": 12.089492608869357,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 30260
    },
    {
      "epoch": 12.089892129444667,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 30261
    },
    {
      "epoch": 12.090291650019976,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 30262
    },
    {
      "epoch": 12.090691170595285,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 30263
    },
    {
      "epoch": 12.091090691170596,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 30264
    },
    {
      "epoch": 12.091490211745905,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 30265
    },
    {
      "epoch": 12.091889732321215,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 30266
    },
    {
      "epoch": 12.092289252896524,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 30267
    },
    {
      "epoch": 12.092688773471833,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 30268
    },
    {
      "epoch": 12.093088294047144,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 30269
    },
    {
      "epoch": 12.093487814622453,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 30270
    },
    {
      "epoch": 12.093887335197763,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 30271
    },
    {
      "epoch": 12.094286855773072,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 30272
    },
    {
      "epoch": 12.094686376348381,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 30273
    },
    {
      "epoch": 12.095085896923692,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 30274
    },
    {
      "epoch": 12.095485417499,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 30275
    },
    {
      "epoch": 12.095884938074311,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 30276
    },
    {
      "epoch": 12.09628445864962,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 30277
    },
    {
      "epoch": 12.09668397922493,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 30278
    },
    {
      "epoch": 12.09708349980024,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0515,
      "step": 30279
    },
    {
      "epoch": 12.097483020375549,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 30280
    },
    {
      "epoch": 12.09788254095086,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 30281
    },
    {
      "epoch": 12.098282061526168,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 30282
    },
    {
      "epoch": 12.098681582101479,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 30283
    },
    {
      "epoch": 12.099081102676788,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 30284
    },
    {
      "epoch": 12.099480623252097,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 30285
    },
    {
      "epoch": 12.099880143827408,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 30286
    },
    {
      "epoch": 12.100279664402716,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 30287
    },
    {
      "epoch": 12.100679184978027,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 30288
    },
    {
      "epoch": 12.101078705553336,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 30289
    },
    {
      "epoch": 12.101478226128645,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 30290
    },
    {
      "epoch": 12.101877746703956,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 30291
    },
    {
      "epoch": 12.102277267279264,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 30292
    },
    {
      "epoch": 12.102676787854575,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 30293
    },
    {
      "epoch": 12.103076308429884,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 30294
    },
    {
      "epoch": 12.103475829005193,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 30295
    },
    {
      "epoch": 12.103875349580504,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 30296
    },
    {
      "epoch": 12.104274870155812,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 30297
    },
    {
      "epoch": 12.104674390731123,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 30298
    },
    {
      "epoch": 12.105073911306432,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 30299
    },
    {
      "epoch": 12.105473431881743,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 30300
    },
    {
      "epoch": 12.105872952457052,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 30301
    },
    {
      "epoch": 12.10627247303236,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 30302
    },
    {
      "epoch": 12.106671993607671,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 30303
    },
    {
      "epoch": 12.10707151418298,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 30304
    },
    {
      "epoch": 12.10747103475829,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 30305
    },
    {
      "epoch": 12.1078705553336,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 30306
    },
    {
      "epoch": 12.108270075908909,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 30307
    },
    {
      "epoch": 12.10866959648422,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 30308
    },
    {
      "epoch": 12.109069117059528,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 30309
    },
    {
      "epoch": 12.109468637634839,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 30310
    },
    {
      "epoch": 12.109868158210148,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 30311
    },
    {
      "epoch": 12.110267678785457,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 30312
    },
    {
      "epoch": 12.110667199360767,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 30313
    },
    {
      "epoch": 12.111066719936076,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 30314
    },
    {
      "epoch": 12.111466240511387,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 30315
    },
    {
      "epoch": 12.111865761086696,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 30316
    },
    {
      "epoch": 12.112265281662006,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 30317
    },
    {
      "epoch": 12.112664802237315,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 30318
    },
    {
      "epoch": 12.113064322812624,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 30319
    },
    {
      "epoch": 12.113463843387935,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 30320
    },
    {
      "epoch": 12.113863363963244,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 30321
    },
    {
      "epoch": 12.114262884538554,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 30322
    },
    {
      "epoch": 12.114662405113863,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 30323
    },
    {
      "epoch": 12.115061925689172,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 30324
    },
    {
      "epoch": 12.115461446264483,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 30325
    },
    {
      "epoch": 12.115860966839792,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 30326
    },
    {
      "epoch": 12.116260487415103,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 30327
    },
    {
      "epoch": 12.116660007990411,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 30328
    },
    {
      "epoch": 12.11705952856572,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 30329
    },
    {
      "epoch": 12.117459049141031,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 30330
    },
    {
      "epoch": 12.11785856971634,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 30331
    },
    {
      "epoch": 12.11825809029165,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 30332
    },
    {
      "epoch": 12.11865761086696,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 30333
    },
    {
      "epoch": 12.11905713144227,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 30334
    },
    {
      "epoch": 12.119456652017579,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 30335
    },
    {
      "epoch": 12.119856172592888,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 30336
    },
    {
      "epoch": 12.120255693168199,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 30337
    },
    {
      "epoch": 12.120655213743508,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 30338
    },
    {
      "epoch": 12.121054734318818,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 30339
    },
    {
      "epoch": 12.121454254894127,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 30340
    },
    {
      "epoch": 12.121853775469436,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 30341
    },
    {
      "epoch": 12.122253296044747,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 30342
    },
    {
      "epoch": 12.122652816620056,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 30343
    },
    {
      "epoch": 12.123052337195366,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 30344
    },
    {
      "epoch": 12.123451857770675,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 30345
    },
    {
      "epoch": 12.123851378345984,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 30346
    },
    {
      "epoch": 12.124250898921295,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 30347
    },
    {
      "epoch": 12.124650419496604,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 30348
    },
    {
      "epoch": 12.125049940071914,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 30349
    },
    {
      "epoch": 12.125449460647223,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 30350
    },
    {
      "epoch": 12.125848981222532,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 30351
    },
    {
      "epoch": 12.126248501797843,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 30352
    },
    {
      "epoch": 12.126648022373152,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 30353
    },
    {
      "epoch": 12.127047542948462,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 30354
    },
    {
      "epoch": 12.127447063523771,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 30355
    },
    {
      "epoch": 12.127846584099082,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 30356
    },
    {
      "epoch": 12.12824610467439,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 30357
    },
    {
      "epoch": 12.1286456252497,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 30358
    },
    {
      "epoch": 12.12904514582501,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 30359
    },
    {
      "epoch": 12.12944466640032,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 30360
    },
    {
      "epoch": 12.12984418697563,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 30361
    },
    {
      "epoch": 12.130243707550939,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 30362
    },
    {
      "epoch": 12.130643228126248,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 30363
    },
    {
      "epoch": 12.131042748701558,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 30364
    },
    {
      "epoch": 12.131442269276867,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 30365
    },
    {
      "epoch": 12.131841789852178,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 30366
    },
    {
      "epoch": 12.132241310427487,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 30367
    },
    {
      "epoch": 12.132640831002796,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 30368
    },
    {
      "epoch": 12.133040351578106,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 30369
    },
    {
      "epoch": 12.133439872153415,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 30370
    },
    {
      "epoch": 12.133839392728726,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 30371
    },
    {
      "epoch": 12.134238913304035,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 30372
    },
    {
      "epoch": 12.134638433879346,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 30373
    },
    {
      "epoch": 12.135037954454654,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 30374
    },
    {
      "epoch": 12.135437475029963,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 30375
    },
    {
      "epoch": 12.135836995605274,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.009,
      "step": 30376
    },
    {
      "epoch": 12.136236516180583,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 30377
    },
    {
      "epoch": 12.136636036755894,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 30378
    },
    {
      "epoch": 12.137035557331203,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 30379
    },
    {
      "epoch": 12.137435077906511,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 30380
    },
    {
      "epoch": 12.137834598481822,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 30381
    },
    {
      "epoch": 12.138234119057131,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 30382
    },
    {
      "epoch": 12.138633639632442,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 30383
    },
    {
      "epoch": 12.13903316020775,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 30384
    },
    {
      "epoch": 12.13943268078306,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 30385
    },
    {
      "epoch": 12.13983220135837,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 30386
    },
    {
      "epoch": 12.140231721933679,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 30387
    },
    {
      "epoch": 12.14063124250899,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 30388
    },
    {
      "epoch": 12.141030763084299,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 30389
    },
    {
      "epoch": 12.14143028365961,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 30390
    },
    {
      "epoch": 12.141829804234918,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 30391
    },
    {
      "epoch": 12.142229324810227,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 30392
    },
    {
      "epoch": 12.142628845385538,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 30393
    },
    {
      "epoch": 12.143028365960847,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 30394
    },
    {
      "epoch": 12.143427886536157,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 30395
    },
    {
      "epoch": 12.143827407111466,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 30396
    },
    {
      "epoch": 12.144226927686775,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 30397
    },
    {
      "epoch": 12.144626448262086,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 30398
    },
    {
      "epoch": 12.145025968837395,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 30399
    },
    {
      "epoch": 12.145425489412705,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 30400
    },
    {
      "epoch": 12.145825009988014,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 30401
    },
    {
      "epoch": 12.146224530563323,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 30402
    },
    {
      "epoch": 12.146624051138634,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 30403
    },
    {
      "epoch": 12.147023571713943,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 30404
    },
    {
      "epoch": 12.147423092289253,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 30405
    },
    {
      "epoch": 12.147822612864562,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 30406
    },
    {
      "epoch": 12.148222133439873,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 30407
    },
    {
      "epoch": 12.148621654015182,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 30408
    },
    {
      "epoch": 12.14902117459049,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 30409
    },
    {
      "epoch": 12.149420695165801,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 30410
    },
    {
      "epoch": 12.14982021574111,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 30411
    },
    {
      "epoch": 12.150219736316421,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 30412
    },
    {
      "epoch": 12.15061925689173,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 30413
    },
    {
      "epoch": 12.151018777467039,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 30414
    },
    {
      "epoch": 12.15141829804235,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 30415
    },
    {
      "epoch": 12.151817818617658,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 30416
    },
    {
      "epoch": 12.152217339192969,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 30417
    },
    {
      "epoch": 12.152616859768278,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 30418
    },
    {
      "epoch": 12.153016380343587,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 30419
    },
    {
      "epoch": 12.153415900918898,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 30420
    },
    {
      "epoch": 12.153815421494206,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 30421
    },
    {
      "epoch": 12.154214942069517,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 30422
    },
    {
      "epoch": 12.154614462644826,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 30423
    },
    {
      "epoch": 12.155013983220137,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 30424
    },
    {
      "epoch": 12.155413503795446,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 30425
    },
    {
      "epoch": 12.155813024370755,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 30426
    },
    {
      "epoch": 12.156212544946065,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 30427
    },
    {
      "epoch": 12.156612065521374,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 30428
    },
    {
      "epoch": 12.157011586096685,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 30429
    },
    {
      "epoch": 12.157411106671994,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 30430
    },
    {
      "epoch": 12.157810627247303,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 30431
    },
    {
      "epoch": 12.158210147822613,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 30432
    },
    {
      "epoch": 12.158609668397922,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 30433
    },
    {
      "epoch": 12.159009188973233,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 30434
    },
    {
      "epoch": 12.159408709548542,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 30435
    },
    {
      "epoch": 12.15980823012385,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 30436
    },
    {
      "epoch": 12.160207750699161,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 30437
    },
    {
      "epoch": 12.16060727127447,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 30438
    },
    {
      "epoch": 12.16100679184978,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 30439
    },
    {
      "epoch": 12.16140631242509,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 30440
    },
    {
      "epoch": 12.1618058330004,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 30441
    },
    {
      "epoch": 12.16220535357571,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 30442
    },
    {
      "epoch": 12.162604874151018,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 30443
    },
    {
      "epoch": 12.163004394726329,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 30444
    },
    {
      "epoch": 12.163403915301638,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 30445
    },
    {
      "epoch": 12.163803435876948,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 30446
    },
    {
      "epoch": 12.164202956452257,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 30447
    },
    {
      "epoch": 12.164602477027566,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 30448
    },
    {
      "epoch": 12.165001997602877,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 30449
    },
    {
      "epoch": 12.165401518178186,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 30450
    },
    {
      "epoch": 12.165801038753496,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 30451
    },
    {
      "epoch": 12.166200559328805,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 30452
    },
    {
      "epoch": 12.166600079904114,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 30453
    },
    {
      "epoch": 12.166999600479425,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 30454
    },
    {
      "epoch": 12.167399121054734,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 30455
    },
    {
      "epoch": 12.167798641630045,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 30456
    },
    {
      "epoch": 12.168198162205353,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 30457
    },
    {
      "epoch": 12.168597682780662,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 30458
    },
    {
      "epoch": 12.168997203355973,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 30459
    },
    {
      "epoch": 12.169396723931282,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 30460
    },
    {
      "epoch": 12.169796244506593,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 30461
    },
    {
      "epoch": 12.170195765081901,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 30462
    },
    {
      "epoch": 12.170595285657212,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 30463
    },
    {
      "epoch": 12.170994806232521,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 30464
    },
    {
      "epoch": 12.17139432680783,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 30465
    },
    {
      "epoch": 12.17179384738314,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 30466
    },
    {
      "epoch": 12.17219336795845,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 30467
    },
    {
      "epoch": 12.17259288853376,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 30468
    },
    {
      "epoch": 12.172992409109069,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 30469
    },
    {
      "epoch": 12.173391929684378,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 30470
    },
    {
      "epoch": 12.173791450259689,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 30471
    },
    {
      "epoch": 12.174190970834998,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 30472
    },
    {
      "epoch": 12.174590491410308,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 30473
    },
    {
      "epoch": 12.174990011985617,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 30474
    },
    {
      "epoch": 12.175389532560926,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 30475
    },
    {
      "epoch": 12.175789053136237,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 30476
    },
    {
      "epoch": 12.176188573711546,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 30477
    },
    {
      "epoch": 12.176588094286856,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 30478
    },
    {
      "epoch": 12.176987614862165,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 30479
    },
    {
      "epoch": 12.177387135437476,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 30480
    },
    {
      "epoch": 12.177786656012785,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 30481
    },
    {
      "epoch": 12.178186176588094,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 30482
    },
    {
      "epoch": 12.178585697163404,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 30483
    },
    {
      "epoch": 12.178985217738713,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 30484
    },
    {
      "epoch": 12.179384738314024,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 30485
    },
    {
      "epoch": 12.179784258889333,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 30486
    },
    {
      "epoch": 12.180183779464642,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 30487
    },
    {
      "epoch": 12.180583300039952,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 30488
    },
    {
      "epoch": 12.180982820615261,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 30489
    },
    {
      "epoch": 12.181382341190572,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 30490
    },
    {
      "epoch": 12.18178186176588,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 30491
    },
    {
      "epoch": 12.18218138234119,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 30492
    },
    {
      "epoch": 12.1825809029165,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 30493
    },
    {
      "epoch": 12.18298042349181,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 30494
    },
    {
      "epoch": 12.18337994406712,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 30495
    },
    {
      "epoch": 12.183779464642429,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 30496
    },
    {
      "epoch": 12.18417898521774,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 30497
    },
    {
      "epoch": 12.184578505793048,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 30498
    },
    {
      "epoch": 12.184978026368357,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 30499
    },
    {
      "epoch": 12.185377546943668,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 30500
    },
    {
      "epoch": 12.185777067518977,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 30501
    },
    {
      "epoch": 12.186176588094288,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 30502
    },
    {
      "epoch": 12.186576108669597,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 30503
    },
    {
      "epoch": 12.186975629244905,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 30504
    },
    {
      "epoch": 12.187375149820216,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 30505
    },
    {
      "epoch": 12.187774670395525,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 30506
    },
    {
      "epoch": 12.188174190970836,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 30507
    },
    {
      "epoch": 12.188573711546145,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 30508
    },
    {
      "epoch": 12.188973232121453,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 30509
    },
    {
      "epoch": 12.189372752696764,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 30510
    },
    {
      "epoch": 12.189772273272073,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 30511
    },
    {
      "epoch": 12.190171793847384,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 30512
    },
    {
      "epoch": 12.190571314422693,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 30513
    },
    {
      "epoch": 12.190970834998003,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 30514
    },
    {
      "epoch": 12.191370355573312,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 30515
    },
    {
      "epoch": 12.191769876148621,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 30516
    },
    {
      "epoch": 12.192169396723932,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 30517
    },
    {
      "epoch": 12.19256891729924,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 30518
    },
    {
      "epoch": 12.192968437874551,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 30519
    },
    {
      "epoch": 12.19336795844986,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 30520
    },
    {
      "epoch": 12.193767479025169,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 30521
    },
    {
      "epoch": 12.19416699960048,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 30522
    },
    {
      "epoch": 12.194566520175789,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 30523
    },
    {
      "epoch": 12.1949660407511,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 30524
    },
    {
      "epoch": 12.195365561326408,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 30525
    },
    {
      "epoch": 12.195765081901717,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 30526
    },
    {
      "epoch": 12.196164602477028,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 30527
    },
    {
      "epoch": 12.196564123052337,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 30528
    },
    {
      "epoch": 12.196963643627647,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 30529
    },
    {
      "epoch": 12.197363164202956,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 30530
    },
    {
      "epoch": 12.197762684778265,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 30531
    },
    {
      "epoch": 12.198162205353576,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 30532
    },
    {
      "epoch": 12.198561725928885,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 30533
    },
    {
      "epoch": 12.198961246504195,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 30534
    },
    {
      "epoch": 12.199360767079504,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 30535
    },
    {
      "epoch": 12.199760287654815,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 30536
    },
    {
      "epoch": 12.200159808230124,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 30537
    },
    {
      "epoch": 12.200559328805433,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 30538
    },
    {
      "epoch": 12.200958849380743,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 30539
    },
    {
      "epoch": 12.201358369956052,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 30540
    },
    {
      "epoch": 12.201757890531363,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 30541
    },
    {
      "epoch": 12.202157411106672,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 30542
    },
    {
      "epoch": 12.20255693168198,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 30543
    },
    {
      "epoch": 12.202956452257292,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 30544
    },
    {
      "epoch": 12.2033559728326,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 30545
    },
    {
      "epoch": 12.203755493407911,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 30546
    },
    {
      "epoch": 12.20415501398322,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 30547
    },
    {
      "epoch": 12.204554534558529,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 30548
    },
    {
      "epoch": 12.20495405513384,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 30549
    },
    {
      "epoch": 12.205353575709148,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 30550
    },
    {
      "epoch": 12.20575309628446,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 30551
    },
    {
      "epoch": 12.206152616859768,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 30552
    },
    {
      "epoch": 12.206552137435079,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 30553
    },
    {
      "epoch": 12.206951658010388,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 30554
    },
    {
      "epoch": 12.207351178585697,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 30555
    },
    {
      "epoch": 12.207750699161007,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 30556
    },
    {
      "epoch": 12.208150219736316,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 30557
    },
    {
      "epoch": 12.208549740311627,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 30558
    },
    {
      "epoch": 12.208949260886936,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 30559
    },
    {
      "epoch": 12.209348781462245,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 30560
    },
    {
      "epoch": 12.209748302037555,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 30561
    },
    {
      "epoch": 12.210147822612864,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 30562
    },
    {
      "epoch": 12.210547343188175,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 30563
    },
    {
      "epoch": 12.210946863763484,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 30564
    },
    {
      "epoch": 12.211346384338793,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 30565
    },
    {
      "epoch": 12.211745904914103,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 30566
    },
    {
      "epoch": 12.212145425489412,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 30567
    },
    {
      "epoch": 12.212544946064723,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 30568
    },
    {
      "epoch": 12.212944466640032,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 30569
    },
    {
      "epoch": 12.213343987215342,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 30570
    },
    {
      "epoch": 12.213743507790651,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 30571
    },
    {
      "epoch": 12.21414302836596,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 30572
    },
    {
      "epoch": 12.214542548941271,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 30573
    },
    {
      "epoch": 12.21494206951658,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 30574
    },
    {
      "epoch": 12.21534159009189,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 30575
    },
    {
      "epoch": 12.2157411106672,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 30576
    },
    {
      "epoch": 12.216140631242508,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 30577
    },
    {
      "epoch": 12.216540151817819,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 30578
    },
    {
      "epoch": 12.216939672393128,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 30579
    },
    {
      "epoch": 12.217339192968439,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 30580
    },
    {
      "epoch": 12.217738713543747,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 30581
    },
    {
      "epoch": 12.218138234119056,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 30582
    },
    {
      "epoch": 12.218537754694367,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 30583
    },
    {
      "epoch": 12.218937275269676,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 30584
    },
    {
      "epoch": 12.219336795844987,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 30585
    },
    {
      "epoch": 12.219736316420295,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 30586
    },
    {
      "epoch": 12.220135836995606,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 30587
    },
    {
      "epoch": 12.220535357570915,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 30588
    },
    {
      "epoch": 12.220934878146224,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 30589
    },
    {
      "epoch": 12.221334398721535,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 30590
    },
    {
      "epoch": 12.221733919296843,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 30591
    },
    {
      "epoch": 12.222133439872154,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 30592
    },
    {
      "epoch": 12.222532960447463,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 30593
    },
    {
      "epoch": 12.222932481022772,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 30594
    },
    {
      "epoch": 12.223332001598083,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 30595
    },
    {
      "epoch": 12.223731522173392,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 30596
    },
    {
      "epoch": 12.224131042748702,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 30597
    },
    {
      "epoch": 12.224530563324011,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 30598
    },
    {
      "epoch": 12.22493008389932,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 30599
    },
    {
      "epoch": 12.22532960447463,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 30600
    },
    {
      "epoch": 12.22572912504994,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 30601
    },
    {
      "epoch": 12.22612864562525,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 30602
    },
    {
      "epoch": 12.22652816620056,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 30603
    },
    {
      "epoch": 12.22692768677587,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 30604
    },
    {
      "epoch": 12.227327207351179,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 30605
    },
    {
      "epoch": 12.227726727926488,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 30606
    },
    {
      "epoch": 12.228126248501798,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 30607
    },
    {
      "epoch": 12.228525769077107,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 30608
    },
    {
      "epoch": 12.228925289652418,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 30609
    },
    {
      "epoch": 12.229324810227727,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 30610
    },
    {
      "epoch": 12.229724330803036,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 30611
    },
    {
      "epoch": 12.230123851378346,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 30612
    },
    {
      "epoch": 12.230523371953655,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 30613
    },
    {
      "epoch": 12.230922892528966,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 30614
    },
    {
      "epoch": 12.231322413104275,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 30615
    },
    {
      "epoch": 12.231721933679584,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 30616
    },
    {
      "epoch": 12.232121454254894,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 30617
    },
    {
      "epoch": 12.232520974830203,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 30618
    },
    {
      "epoch": 12.232920495405514,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 30619
    },
    {
      "epoch": 12.233320015980823,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 30620
    },
    {
      "epoch": 12.233719536556134,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 30621
    },
    {
      "epoch": 12.234119057131442,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 30622
    },
    {
      "epoch": 12.234518577706751,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 30623
    },
    {
      "epoch": 12.234918098282062,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 30624
    },
    {
      "epoch": 12.235317618857371,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 30625
    },
    {
      "epoch": 12.235717139432682,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 30626
    },
    {
      "epoch": 12.23611666000799,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 30627
    },
    {
      "epoch": 12.2365161805833,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 30628
    },
    {
      "epoch": 12.23691570115861,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 30629
    },
    {
      "epoch": 12.237315221733919,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 30630
    },
    {
      "epoch": 12.23771474230923,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 30631
    },
    {
      "epoch": 12.238114262884539,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 30632
    },
    {
      "epoch": 12.238513783459847,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 30633
    },
    {
      "epoch": 12.238913304035158,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 30634
    },
    {
      "epoch": 12.239312824610467,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 30635
    },
    {
      "epoch": 12.239712345185778,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 30636
    },
    {
      "epoch": 12.240111865761087,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 30637
    },
    {
      "epoch": 12.240511386336395,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 30638
    },
    {
      "epoch": 12.240910906911706,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 30639
    },
    {
      "epoch": 12.241310427487015,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 30640
    },
    {
      "epoch": 12.241709948062326,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 30641
    },
    {
      "epoch": 12.242109468637635,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 30642
    },
    {
      "epoch": 12.242508989212945,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 30643
    },
    {
      "epoch": 12.242908509788254,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 30644
    },
    {
      "epoch": 12.243308030363563,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 30645
    },
    {
      "epoch": 12.243707550938874,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 30646
    },
    {
      "epoch": 12.244107071514183,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 30647
    },
    {
      "epoch": 12.244506592089493,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 30648
    },
    {
      "epoch": 12.244906112664802,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 30649
    },
    {
      "epoch": 12.245305633240111,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 30650
    },
    {
      "epoch": 12.245705153815422,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 30651
    },
    {
      "epoch": 12.24610467439073,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 30652
    },
    {
      "epoch": 12.246504194966041,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 30653
    },
    {
      "epoch": 12.24690371554135,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 30654
    },
    {
      "epoch": 12.24730323611666,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 30655
    },
    {
      "epoch": 12.24770275669197,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 30656
    },
    {
      "epoch": 12.248102277267279,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 30657
    },
    {
      "epoch": 12.24850179784259,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 30658
    },
    {
      "epoch": 12.248901318417898,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 30659
    },
    {
      "epoch": 12.249300838993209,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 30660
    },
    {
      "epoch": 12.249700359568518,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 30661
    },
    {
      "epoch": 12.250099880143827,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 30662
    },
    {
      "epoch": 12.250499400719137,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 30663
    },
    {
      "epoch": 12.250898921294446,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 30664
    },
    {
      "epoch": 12.251298441869757,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 30665
    },
    {
      "epoch": 12.251697962445066,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 30666
    },
    {
      "epoch": 12.252097483020375,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 30667
    },
    {
      "epoch": 12.252497003595685,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 30668
    },
    {
      "epoch": 12.252896524170994,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 30669
    },
    {
      "epoch": 12.253296044746305,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 30670
    },
    {
      "epoch": 12.253695565321614,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 30671
    },
    {
      "epoch": 12.254095085896923,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 30672
    },
    {
      "epoch": 12.254494606472234,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 30673
    },
    {
      "epoch": 12.254894127047542,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 30674
    },
    {
      "epoch": 12.255293647622853,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 30675
    },
    {
      "epoch": 12.255693168198162,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 30676
    },
    {
      "epoch": 12.256092688773473,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 30677
    },
    {
      "epoch": 12.256492209348782,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 30678
    },
    {
      "epoch": 12.25689172992409,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 30679
    },
    {
      "epoch": 12.257291250499401,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 30680
    },
    {
      "epoch": 12.25769077107471,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 30681
    },
    {
      "epoch": 12.25809029165002,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 30682
    },
    {
      "epoch": 12.25848981222533,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 30683
    },
    {
      "epoch": 12.258889332800639,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 30684
    },
    {
      "epoch": 12.25928885337595,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 30685
    },
    {
      "epoch": 12.259688373951258,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 30686
    },
    {
      "epoch": 12.260087894526569,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 30687
    },
    {
      "epoch": 12.260487415101878,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 30688
    },
    {
      "epoch": 12.260886935677187,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 30689
    },
    {
      "epoch": 12.261286456252497,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 30690
    },
    {
      "epoch": 12.261685976827806,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 30691
    },
    {
      "epoch": 12.262085497403117,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 30692
    },
    {
      "epoch": 12.262485017978426,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 30693
    },
    {
      "epoch": 12.262884538553735,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 30694
    },
    {
      "epoch": 12.263284059129045,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 30695
    },
    {
      "epoch": 12.263683579704354,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 30696
    },
    {
      "epoch": 12.264083100279665,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 30697
    },
    {
      "epoch": 12.264482620854974,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 30698
    },
    {
      "epoch": 12.264882141430284,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 30699
    },
    {
      "epoch": 12.265281662005593,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 30700
    },
    {
      "epoch": 12.265681182580902,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 30701
    },
    {
      "epoch": 12.266080703156213,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 30702
    },
    {
      "epoch": 12.266480223731522,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 30703
    },
    {
      "epoch": 12.266879744306832,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 30704
    },
    {
      "epoch": 12.267279264882141,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 30705
    },
    {
      "epoch": 12.26767878545745,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 30706
    },
    {
      "epoch": 12.268078306032761,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 30707
    },
    {
      "epoch": 12.26847782660807,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 30708
    },
    {
      "epoch": 12.26887734718338,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 30709
    },
    {
      "epoch": 12.26927686775869,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 30710
    },
    {
      "epoch": 12.269676388333998,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 30711
    },
    {
      "epoch": 12.270075908909309,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 30712
    },
    {
      "epoch": 12.270475429484618,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 30713
    },
    {
      "epoch": 12.270874950059929,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 30714
    },
    {
      "epoch": 12.271274470635237,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 30715
    },
    {
      "epoch": 12.271673991210548,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 30716
    },
    {
      "epoch": 12.272073511785857,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 30717
    },
    {
      "epoch": 12.272473032361166,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 30718
    },
    {
      "epoch": 12.272872552936477,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 30719
    },
    {
      "epoch": 12.273272073511786,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 30720
    },
    {
      "epoch": 12.273671594087096,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 30721
    },
    {
      "epoch": 12.274071114662405,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 30722
    },
    {
      "epoch": 12.274470635237714,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 30723
    },
    {
      "epoch": 12.274870155813025,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 30724
    },
    {
      "epoch": 12.275269676388334,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 30725
    },
    {
      "epoch": 12.275669196963644,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 30726
    },
    {
      "epoch": 12.276068717538953,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 30727
    },
    {
      "epoch": 12.276468238114262,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 30728
    },
    {
      "epoch": 12.276867758689573,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 30729
    },
    {
      "epoch": 12.277267279264882,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 30730
    },
    {
      "epoch": 12.277666799840192,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 30731
    },
    {
      "epoch": 12.278066320415501,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 30732
    },
    {
      "epoch": 12.278465840990812,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 30733
    },
    {
      "epoch": 12.27886536156612,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 30734
    },
    {
      "epoch": 12.27926488214143,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 30735
    },
    {
      "epoch": 12.27966440271674,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 30736
    },
    {
      "epoch": 12.28006392329205,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 30737
    },
    {
      "epoch": 12.28046344386736,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 30738
    },
    {
      "epoch": 12.280862964442669,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 30739
    },
    {
      "epoch": 12.281262485017978,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 30740
    },
    {
      "epoch": 12.281662005593288,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 30741
    },
    {
      "epoch": 12.282061526168597,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 30742
    },
    {
      "epoch": 12.282461046743908,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 30743
    },
    {
      "epoch": 12.282860567319217,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 30744
    },
    {
      "epoch": 12.283260087894526,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 30745
    },
    {
      "epoch": 12.283659608469836,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 30746
    },
    {
      "epoch": 12.284059129045145,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0077,
      "step": 30747
    },
    {
      "epoch": 12.284458649620456,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 30748
    },
    {
      "epoch": 12.284858170195765,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 30749
    },
    {
      "epoch": 12.285257690771076,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 30750
    },
    {
      "epoch": 12.285657211346384,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 30751
    },
    {
      "epoch": 12.286056731921693,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0128,
      "step": 30752
    },
    {
      "epoch": 12.286456252497004,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 30753
    },
    {
      "epoch": 12.286855773072313,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 30754
    },
    {
      "epoch": 12.287255293647624,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 30755
    },
    {
      "epoch": 12.287654814222932,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 30756
    },
    {
      "epoch": 12.288054334798241,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 30757
    },
    {
      "epoch": 12.288453855373552,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 30758
    },
    {
      "epoch": 12.288853375948861,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 30759
    },
    {
      "epoch": 12.289252896524172,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 30760
    },
    {
      "epoch": 12.28965241709948,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 30761
    },
    {
      "epoch": 12.29005193767479,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 30762
    },
    {
      "epoch": 12.2904514582501,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 30763
    },
    {
      "epoch": 12.290850978825409,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 30764
    },
    {
      "epoch": 12.29125049940072,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 30765
    },
    {
      "epoch": 12.291650019976029,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 30766
    },
    {
      "epoch": 12.29204954055134,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 30767
    },
    {
      "epoch": 12.292449061126648,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 30768
    },
    {
      "epoch": 12.292848581701957,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 30769
    },
    {
      "epoch": 12.293248102277268,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 30770
    },
    {
      "epoch": 12.293647622852577,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 30771
    },
    {
      "epoch": 12.294047143427887,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 30772
    },
    {
      "epoch": 12.294446664003196,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 30773
    },
    {
      "epoch": 12.294846184578505,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 30774
    },
    {
      "epoch": 12.295245705153816,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 30775
    },
    {
      "epoch": 12.295645225729125,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 30776
    },
    {
      "epoch": 12.296044746304435,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 30777
    },
    {
      "epoch": 12.296444266879744,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 30778
    },
    {
      "epoch": 12.296843787455053,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 30779
    },
    {
      "epoch": 12.297243308030364,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 30780
    },
    {
      "epoch": 12.297642828605673,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 30781
    },
    {
      "epoch": 12.298042349180983,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 30782
    },
    {
      "epoch": 12.298441869756292,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 30783
    },
    {
      "epoch": 12.298841390331603,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 30784
    },
    {
      "epoch": 12.299240910906912,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 30785
    },
    {
      "epoch": 12.29964043148222,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 30786
    },
    {
      "epoch": 12.300039952057531,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 30787
    },
    {
      "epoch": 12.30043947263284,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 30788
    },
    {
      "epoch": 12.300838993208151,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 30789
    },
    {
      "epoch": 12.30123851378346,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 30790
    },
    {
      "epoch": 12.301638034358769,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 30791
    },
    {
      "epoch": 12.30203755493408,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 30792
    },
    {
      "epoch": 12.302437075509388,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 30793
    },
    {
      "epoch": 12.302836596084699,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 30794
    },
    {
      "epoch": 12.303236116660008,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 30795
    },
    {
      "epoch": 12.303635637235317,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 30796
    },
    {
      "epoch": 12.304035157810628,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 30797
    },
    {
      "epoch": 12.304434678385936,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 30798
    },
    {
      "epoch": 12.304834198961247,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 30799
    },
    {
      "epoch": 12.305233719536556,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 30800
    },
    {
      "epoch": 12.305633240111867,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 30801
    },
    {
      "epoch": 12.306032760687176,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 30802
    },
    {
      "epoch": 12.306432281262484,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 30803
    },
    {
      "epoch": 12.306831801837795,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 30804
    },
    {
      "epoch": 12.307231322413104,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 30805
    },
    {
      "epoch": 12.307630842988415,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 30806
    },
    {
      "epoch": 12.308030363563724,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 30807
    },
    {
      "epoch": 12.308429884139032,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 30808
    },
    {
      "epoch": 12.308829404714343,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 30809
    },
    {
      "epoch": 12.309228925289652,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 30810
    },
    {
      "epoch": 12.309628445864963,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 30811
    },
    {
      "epoch": 12.310027966440272,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 30812
    },
    {
      "epoch": 12.31042748701558,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 30813
    },
    {
      "epoch": 12.310827007590891,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 30814
    },
    {
      "epoch": 12.3112265281662,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 30815
    },
    {
      "epoch": 12.31162604874151,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 30816
    },
    {
      "epoch": 12.31202556931682,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 30817
    },
    {
      "epoch": 12.31242508989213,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 30818
    },
    {
      "epoch": 12.31282461046744,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 30819
    },
    {
      "epoch": 12.313224131042748,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 30820
    },
    {
      "epoch": 12.313623651618059,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 30821
    },
    {
      "epoch": 12.314023172193368,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 30822
    },
    {
      "epoch": 12.314422692768678,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 30823
    },
    {
      "epoch": 12.314822213343987,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 30824
    },
    {
      "epoch": 12.315221733919296,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 30825
    },
    {
      "epoch": 12.315621254494607,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 30826
    },
    {
      "epoch": 12.316020775069916,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 30827
    },
    {
      "epoch": 12.316420295645226,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 30828
    },
    {
      "epoch": 12.316819816220535,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 30829
    },
    {
      "epoch": 12.317219336795844,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 30830
    },
    {
      "epoch": 12.317618857371155,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 30831
    },
    {
      "epoch": 12.318018377946464,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 30832
    },
    {
      "epoch": 12.318417898521774,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 30833
    },
    {
      "epoch": 12.318817419097083,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 30834
    },
    {
      "epoch": 12.319216939672392,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 30835
    },
    {
      "epoch": 12.319616460247703,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 30836
    },
    {
      "epoch": 12.320015980823012,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 30837
    },
    {
      "epoch": 12.320415501398323,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 30838
    },
    {
      "epoch": 12.320815021973631,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 30839
    },
    {
      "epoch": 12.321214542548942,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 30840
    },
    {
      "epoch": 12.321614063124251,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 30841
    },
    {
      "epoch": 12.32201358369956,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 30842
    },
    {
      "epoch": 12.32241310427487,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 30843
    },
    {
      "epoch": 12.32281262485018,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 30844
    },
    {
      "epoch": 12.32321214542549,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 30845
    },
    {
      "epoch": 12.323611666000799,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 30846
    },
    {
      "epoch": 12.324011186576108,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 30847
    },
    {
      "epoch": 12.324410707151419,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 30848
    },
    {
      "epoch": 12.324810227726728,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 30849
    },
    {
      "epoch": 12.325209748302038,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 30850
    },
    {
      "epoch": 12.325609268877347,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 30851
    },
    {
      "epoch": 12.326008789452656,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 30852
    },
    {
      "epoch": 12.326408310027967,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 30853
    },
    {
      "epoch": 12.326807830603276,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 30854
    },
    {
      "epoch": 12.327207351178586,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 30855
    },
    {
      "epoch": 12.327606871753895,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 30856
    },
    {
      "epoch": 12.328006392329206,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 30857
    },
    {
      "epoch": 12.328405912904515,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 30858
    },
    {
      "epoch": 12.328805433479824,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 30859
    },
    {
      "epoch": 12.329204954055134,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 30860
    },
    {
      "epoch": 12.329604474630443,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 30861
    },
    {
      "epoch": 12.330003995205754,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 30862
    },
    {
      "epoch": 12.330403515781063,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 30863
    },
    {
      "epoch": 12.330803036356372,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 30864
    },
    {
      "epoch": 12.331202556931682,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 30865
    },
    {
      "epoch": 12.331602077506991,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 30866
    },
    {
      "epoch": 12.332001598082302,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 30867
    },
    {
      "epoch": 12.33240111865761,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 30868
    },
    {
      "epoch": 12.33280063923292,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 30869
    },
    {
      "epoch": 12.33320015980823,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 30870
    },
    {
      "epoch": 12.33359968038354,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 30871
    },
    {
      "epoch": 12.33399920095885,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 30872
    },
    {
      "epoch": 12.334398721534159,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 30873
    },
    {
      "epoch": 12.334798242109468,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 30874
    },
    {
      "epoch": 12.335197762684778,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 30875
    },
    {
      "epoch": 12.335597283260087,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 30876
    },
    {
      "epoch": 12.335996803835398,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 30877
    },
    {
      "epoch": 12.336396324410707,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 30878
    },
    {
      "epoch": 12.336795844986018,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 30879
    },
    {
      "epoch": 12.337195365561326,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 30880
    },
    {
      "epoch": 12.337594886136635,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 30881
    },
    {
      "epoch": 12.337994406711946,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 30882
    },
    {
      "epoch": 12.338393927287255,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 30883
    },
    {
      "epoch": 12.338793447862566,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 30884
    },
    {
      "epoch": 12.339192968437874,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 30885
    },
    {
      "epoch": 12.339592489013183,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 30886
    },
    {
      "epoch": 12.339992009588494,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 30887
    },
    {
      "epoch": 12.340391530163803,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 30888
    },
    {
      "epoch": 12.340791050739114,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 30889
    },
    {
      "epoch": 12.341190571314423,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 30890
    },
    {
      "epoch": 12.341590091889731,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 30891
    },
    {
      "epoch": 12.341989612465042,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 30892
    },
    {
      "epoch": 12.342389133040351,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 30893
    },
    {
      "epoch": 12.342788653615662,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 30894
    },
    {
      "epoch": 12.34318817419097,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 30895
    },
    {
      "epoch": 12.343587694766281,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 30896
    },
    {
      "epoch": 12.34398721534159,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 30897
    },
    {
      "epoch": 12.344386735916899,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 30898
    },
    {
      "epoch": 12.34478625649221,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 30899
    },
    {
      "epoch": 12.345185777067519,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 30900
    },
    {
      "epoch": 12.34558529764283,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 30901
    },
    {
      "epoch": 12.345984818218138,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 30902
    },
    {
      "epoch": 12.346384338793447,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 30903
    },
    {
      "epoch": 12.346783859368758,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 30904
    },
    {
      "epoch": 12.347183379944067,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 30905
    },
    {
      "epoch": 12.347582900519377,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 30906
    },
    {
      "epoch": 12.347982421094686,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 30907
    },
    {
      "epoch": 12.348381941669995,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 30908
    },
    {
      "epoch": 12.348781462245306,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 30909
    },
    {
      "epoch": 12.349180982820615,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 30910
    },
    {
      "epoch": 12.349580503395925,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 30911
    },
    {
      "epoch": 12.349980023971234,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 30912
    },
    {
      "epoch": 12.350379544546545,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 30913
    },
    {
      "epoch": 12.350779065121854,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 30914
    },
    {
      "epoch": 12.351178585697163,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 30915
    },
    {
      "epoch": 12.351578106272473,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 30916
    },
    {
      "epoch": 12.351977626847782,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 30917
    },
    {
      "epoch": 12.352377147423093,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 30918
    },
    {
      "epoch": 12.352776667998402,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 30919
    },
    {
      "epoch": 12.35317618857371,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 30920
    },
    {
      "epoch": 12.353575709149021,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 30921
    },
    {
      "epoch": 12.35397522972433,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0066,
      "step": 30922
    },
    {
      "epoch": 12.354374750299641,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 30923
    },
    {
      "epoch": 12.35477427087495,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 30924
    },
    {
      "epoch": 12.355173791450259,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 30925
    },
    {
      "epoch": 12.35557331202557,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 30926
    },
    {
      "epoch": 12.355972832600878,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 30927
    },
    {
      "epoch": 12.356372353176189,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 30928
    },
    {
      "epoch": 12.356771873751498,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 30929
    },
    {
      "epoch": 12.357171394326809,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 30930
    },
    {
      "epoch": 12.357570914902118,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 30931
    },
    {
      "epoch": 12.357970435477426,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 30932
    },
    {
      "epoch": 12.358369956052737,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 30933
    },
    {
      "epoch": 12.358769476628046,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 30934
    },
    {
      "epoch": 12.359168997203357,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 30935
    },
    {
      "epoch": 12.359568517778666,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 30936
    },
    {
      "epoch": 12.359968038353975,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 30937
    },
    {
      "epoch": 12.360367558929285,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 30938
    },
    {
      "epoch": 12.360767079504594,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 30939
    },
    {
      "epoch": 12.361166600079905,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 30940
    },
    {
      "epoch": 12.361566120655214,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 30941
    },
    {
      "epoch": 12.361965641230523,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 30942
    },
    {
      "epoch": 12.362365161805833,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 30943
    },
    {
      "epoch": 12.362764682381142,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 30944
    },
    {
      "epoch": 12.363164202956453,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 30945
    },
    {
      "epoch": 12.363563723531762,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 30946
    },
    {
      "epoch": 12.363963244107072,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 30947
    },
    {
      "epoch": 12.364362764682381,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 30948
    },
    {
      "epoch": 12.36476228525769,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 30949
    },
    {
      "epoch": 12.365161805833,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 30950
    },
    {
      "epoch": 12.36556132640831,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 30951
    },
    {
      "epoch": 12.36596084698362,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 30952
    },
    {
      "epoch": 12.36636036755893,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 30953
    },
    {
      "epoch": 12.366759888134238,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 30954
    },
    {
      "epoch": 12.367159408709549,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 30955
    },
    {
      "epoch": 12.367558929284858,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 30956
    },
    {
      "epoch": 12.367958449860168,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 30957
    },
    {
      "epoch": 12.368357970435477,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 30958
    },
    {
      "epoch": 12.368757491010786,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 30959
    },
    {
      "epoch": 12.369157011586097,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 30960
    },
    {
      "epoch": 12.369556532161406,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 30961
    },
    {
      "epoch": 12.369956052736717,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 30962
    },
    {
      "epoch": 12.370355573312025,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 30963
    },
    {
      "epoch": 12.370755093887336,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 30964
    },
    {
      "epoch": 12.371154614462645,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 30965
    },
    {
      "epoch": 12.371554135037954,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 30966
    },
    {
      "epoch": 12.371953655613265,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0097,
      "step": 30967
    },
    {
      "epoch": 12.372353176188573,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 30968
    },
    {
      "epoch": 12.372752696763884,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 30969
    },
    {
      "epoch": 12.373152217339193,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 30970
    },
    {
      "epoch": 12.373551737914502,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 30971
    },
    {
      "epoch": 12.373951258489813,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 30972
    },
    {
      "epoch": 12.374350779065121,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 30973
    },
    {
      "epoch": 12.374750299640432,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 30974
    },
    {
      "epoch": 12.375149820215741,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 30975
    },
    {
      "epoch": 12.37554934079105,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 30976
    },
    {
      "epoch": 12.37594886136636,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 30977
    },
    {
      "epoch": 12.37634838194167,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 30978
    },
    {
      "epoch": 12.37674790251698,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 30979
    },
    {
      "epoch": 12.377147423092289,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 30980
    },
    {
      "epoch": 12.3775469436676,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 30981
    },
    {
      "epoch": 12.377946464242909,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 30982
    },
    {
      "epoch": 12.378345984818218,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 30983
    },
    {
      "epoch": 12.378745505393528,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 30984
    },
    {
      "epoch": 12.379145025968837,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 30985
    },
    {
      "epoch": 12.379544546544148,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 30986
    },
    {
      "epoch": 12.379944067119457,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 30987
    },
    {
      "epoch": 12.380343587694766,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 30988
    },
    {
      "epoch": 12.380743108270076,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 30989
    },
    {
      "epoch": 12.381142628845385,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 30990
    },
    {
      "epoch": 12.381542149420696,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 30991
    },
    {
      "epoch": 12.381941669996005,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 30992
    },
    {
      "epoch": 12.382341190571314,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 30993
    },
    {
      "epoch": 12.382740711146624,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 30994
    },
    {
      "epoch": 12.383140231721933,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 30995
    },
    {
      "epoch": 12.383539752297244,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 30996
    },
    {
      "epoch": 12.383939272872553,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 30997
    },
    {
      "epoch": 12.384338793447863,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 30998
    },
    {
      "epoch": 12.384738314023172,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 30999
    },
    {
      "epoch": 12.385137834598481,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 31000
    },
    {
      "epoch": 12.385537355173792,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 31001
    },
    {
      "epoch": 12.3859368757491,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 31002
    },
    {
      "epoch": 12.386336396324412,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 31003
    },
    {
      "epoch": 12.38673591689972,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 31004
    },
    {
      "epoch": 12.38713543747503,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 31005
    },
    {
      "epoch": 12.38753495805034,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 31006
    },
    {
      "epoch": 12.387934478625649,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 31007
    },
    {
      "epoch": 12.38833399920096,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 31008
    },
    {
      "epoch": 12.388733519776268,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 31009
    },
    {
      "epoch": 12.389133040351577,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 31010
    },
    {
      "epoch": 12.389532560926888,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 31011
    },
    {
      "epoch": 12.389932081502197,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 31012
    },
    {
      "epoch": 12.390331602077508,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 31013
    },
    {
      "epoch": 12.390731122652817,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 31014
    },
    {
      "epoch": 12.391130643228125,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0117,
      "step": 31015
    },
    {
      "epoch": 12.391530163803436,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 31016
    },
    {
      "epoch": 12.391929684378745,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 31017
    },
    {
      "epoch": 12.392329204954056,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 31018
    },
    {
      "epoch": 12.392728725529365,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 31019
    },
    {
      "epoch": 12.393128246104675,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 31020
    },
    {
      "epoch": 12.393527766679984,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 31021
    },
    {
      "epoch": 12.393927287255293,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 31022
    },
    {
      "epoch": 12.394326807830604,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 31023
    },
    {
      "epoch": 12.394726328405913,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 31024
    },
    {
      "epoch": 12.395125848981223,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 31025
    },
    {
      "epoch": 12.395525369556532,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 31026
    },
    {
      "epoch": 12.395924890131841,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 31027
    },
    {
      "epoch": 12.396324410707152,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 31028
    },
    {
      "epoch": 12.39672393128246,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 31029
    },
    {
      "epoch": 12.397123451857771,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 31030
    },
    {
      "epoch": 12.39752297243308,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 31031
    },
    {
      "epoch": 12.397922493008389,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 31032
    },
    {
      "epoch": 12.3983220135837,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 31033
    },
    {
      "epoch": 12.398721534159009,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 31034
    },
    {
      "epoch": 12.39912105473432,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 31035
    },
    {
      "epoch": 12.399520575309628,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 31036
    },
    {
      "epoch": 12.399920095884939,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 31037
    },
    {
      "epoch": 12.400319616460248,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 31038
    },
    {
      "epoch": 12.400719137035557,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 31039
    },
    {
      "epoch": 12.401118657610867,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 31040
    },
    {
      "epoch": 12.401518178186176,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 31041
    },
    {
      "epoch": 12.401917698761487,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 31042
    },
    {
      "epoch": 12.402317219336796,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 31043
    },
    {
      "epoch": 12.402716739912105,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 31044
    },
    {
      "epoch": 12.403116260487415,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 31045
    },
    {
      "epoch": 12.403515781062724,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 31046
    },
    {
      "epoch": 12.403915301638035,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 31047
    },
    {
      "epoch": 12.404314822213344,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 31048
    },
    {
      "epoch": 12.404714342788653,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 31049
    },
    {
      "epoch": 12.405113863363963,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 31050
    },
    {
      "epoch": 12.405513383939272,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 31051
    },
    {
      "epoch": 12.405912904514583,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 31052
    },
    {
      "epoch": 12.406312425089892,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 31053
    },
    {
      "epoch": 12.4067119456652,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 31054
    },
    {
      "epoch": 12.407111466240512,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 31055
    },
    {
      "epoch": 12.40751098681582,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 31056
    },
    {
      "epoch": 12.407910507391131,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 31057
    },
    {
      "epoch": 12.40831002796644,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 31058
    },
    {
      "epoch": 12.40870954854175,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 31059
    },
    {
      "epoch": 12.40910906911706,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 31060
    },
    {
      "epoch": 12.409508589692368,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 31061
    },
    {
      "epoch": 12.40990811026768,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 31062
    },
    {
      "epoch": 12.410307630842988,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 31063
    },
    {
      "epoch": 12.410707151418299,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 31064
    },
    {
      "epoch": 12.411106671993608,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 31065
    },
    {
      "epoch": 12.411506192568917,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 31066
    },
    {
      "epoch": 12.411905713144227,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 31067
    },
    {
      "epoch": 12.412305233719536,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 31068
    },
    {
      "epoch": 12.412704754294847,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 31069
    },
    {
      "epoch": 12.413104274870156,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 31070
    },
    {
      "epoch": 12.413503795445465,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 31071
    },
    {
      "epoch": 12.413903316020775,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 31072
    },
    {
      "epoch": 12.414302836596084,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 31073
    },
    {
      "epoch": 12.414702357171395,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 31074
    },
    {
      "epoch": 12.415101877746704,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 31075
    },
    {
      "epoch": 12.415501398322014,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 31076
    },
    {
      "epoch": 12.415900918897323,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 31077
    },
    {
      "epoch": 12.416300439472632,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 31078
    },
    {
      "epoch": 12.416699960047943,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 31079
    },
    {
      "epoch": 12.417099480623252,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 31080
    },
    {
      "epoch": 12.417499001198562,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 31081
    },
    {
      "epoch": 12.417898521773871,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 31082
    },
    {
      "epoch": 12.41829804234918,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 31083
    },
    {
      "epoch": 12.418697562924491,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 31084
    },
    {
      "epoch": 12.4190970834998,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 31085
    },
    {
      "epoch": 12.41949660407511,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 31086
    },
    {
      "epoch": 12.41989612465042,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 31087
    },
    {
      "epoch": 12.420295645225728,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 31088
    },
    {
      "epoch": 12.420695165801039,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 31089
    },
    {
      "epoch": 12.421094686376348,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 31090
    },
    {
      "epoch": 12.421494206951659,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 31091
    },
    {
      "epoch": 12.421893727526967,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 31092
    },
    {
      "epoch": 12.422293248102278,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 31093
    },
    {
      "epoch": 12.422692768677587,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 31094
    },
    {
      "epoch": 12.423092289252896,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 31095
    },
    {
      "epoch": 12.423491809828207,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 31096
    },
    {
      "epoch": 12.423891330403515,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 31097
    },
    {
      "epoch": 12.424290850978826,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 31098
    },
    {
      "epoch": 12.424690371554135,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 31099
    },
    {
      "epoch": 12.425089892129444,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 31100
    },
    {
      "epoch": 12.425489412704755,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 31101
    },
    {
      "epoch": 12.425888933280063,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 31102
    },
    {
      "epoch": 12.426288453855374,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 31103
    },
    {
      "epoch": 12.426687974430683,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 31104
    },
    {
      "epoch": 12.427087495005992,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 31105
    },
    {
      "epoch": 12.427487015581303,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 31106
    },
    {
      "epoch": 12.427886536156612,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 31107
    },
    {
      "epoch": 12.428286056731922,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 31108
    },
    {
      "epoch": 12.428685577307231,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 31109
    },
    {
      "epoch": 12.429085097882542,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 31110
    },
    {
      "epoch": 12.42948461845785,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 31111
    },
    {
      "epoch": 12.42988413903316,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 31112
    },
    {
      "epoch": 12.43028365960847,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 31113
    },
    {
      "epoch": 12.43068318018378,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 31114
    },
    {
      "epoch": 12.43108270075909,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 31115
    },
    {
      "epoch": 12.431482221334399,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 31116
    },
    {
      "epoch": 12.431881741909708,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 31117
    },
    {
      "epoch": 12.432281262485018,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 31118
    },
    {
      "epoch": 12.432680783060327,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 31119
    },
    {
      "epoch": 12.433080303635638,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 31120
    },
    {
      "epoch": 12.433479824210947,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 31121
    },
    {
      "epoch": 12.433879344786256,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 31122
    },
    {
      "epoch": 12.434278865361566,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 31123
    },
    {
      "epoch": 12.434678385936875,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 31124
    },
    {
      "epoch": 12.435077906512186,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 31125
    },
    {
      "epoch": 12.435477427087495,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 31126
    },
    {
      "epoch": 12.435876947662805,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 31127
    },
    {
      "epoch": 12.436276468238114,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 31128
    },
    {
      "epoch": 12.436675988813423,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 31129
    },
    {
      "epoch": 12.437075509388734,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 31130
    },
    {
      "epoch": 12.437475029964043,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 31131
    },
    {
      "epoch": 12.437874550539354,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 31132
    },
    {
      "epoch": 12.438274071114662,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 31133
    },
    {
      "epoch": 12.438673591689971,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 31134
    },
    {
      "epoch": 12.439073112265282,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 31135
    },
    {
      "epoch": 12.439472632840591,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 31136
    },
    {
      "epoch": 12.439872153415902,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 31137
    },
    {
      "epoch": 12.44027167399121,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 31138
    },
    {
      "epoch": 12.44067119456652,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 31139
    },
    {
      "epoch": 12.44107071514183,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 31140
    },
    {
      "epoch": 12.441470235717139,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 31141
    },
    {
      "epoch": 12.44186975629245,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 31142
    },
    {
      "epoch": 12.442269276867759,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 31143
    },
    {
      "epoch": 12.44266879744307,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 31144
    },
    {
      "epoch": 12.443068318018378,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 31145
    },
    {
      "epoch": 12.443467838593687,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 31146
    },
    {
      "epoch": 12.443867359168998,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 31147
    },
    {
      "epoch": 12.444266879744307,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 31148
    },
    {
      "epoch": 12.444666400319617,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 31149
    },
    {
      "epoch": 12.445065920894926,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 31150
    },
    {
      "epoch": 12.445465441470235,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 31151
    },
    {
      "epoch": 12.445864962045546,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 31152
    },
    {
      "epoch": 12.446264482620855,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 31153
    },
    {
      "epoch": 12.446664003196165,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 31154
    },
    {
      "epoch": 12.447063523771474,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 31155
    },
    {
      "epoch": 12.447463044346783,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 31156
    },
    {
      "epoch": 12.447862564922094,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 31157
    },
    {
      "epoch": 12.448262085497403,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 31158
    },
    {
      "epoch": 12.448661606072713,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 31159
    },
    {
      "epoch": 12.449061126648022,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 31160
    },
    {
      "epoch": 12.449460647223333,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 31161
    },
    {
      "epoch": 12.449860167798642,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 31162
    },
    {
      "epoch": 12.45025968837395,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 31163
    },
    {
      "epoch": 12.450659208949261,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 31164
    },
    {
      "epoch": 12.45105872952457,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 31165
    },
    {
      "epoch": 12.451458250099881,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 31166
    },
    {
      "epoch": 12.45185777067519,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 31167
    },
    {
      "epoch": 12.452257291250499,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 31168
    },
    {
      "epoch": 12.45265681182581,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 31169
    },
    {
      "epoch": 12.453056332401118,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0135,
      "step": 31170
    },
    {
      "epoch": 12.453455852976429,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 31171
    },
    {
      "epoch": 12.453855373551738,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 31172
    },
    {
      "epoch": 12.454254894127047,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 31173
    },
    {
      "epoch": 12.454654414702357,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 31174
    },
    {
      "epoch": 12.455053935277666,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 31175
    },
    {
      "epoch": 12.455453455852977,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 31176
    },
    {
      "epoch": 12.455852976428286,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 31177
    },
    {
      "epoch": 12.456252497003595,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 31178
    },
    {
      "epoch": 12.456652017578906,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 31179
    },
    {
      "epoch": 12.457051538154214,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 31180
    },
    {
      "epoch": 12.457451058729525,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 31181
    },
    {
      "epoch": 12.457850579304834,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 31182
    },
    {
      "epoch": 12.458250099880145,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 31183
    },
    {
      "epoch": 12.458649620455454,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 31184
    },
    {
      "epoch": 12.459049141030762,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 31185
    },
    {
      "epoch": 12.459448661606073,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 31186
    },
    {
      "epoch": 12.459848182181382,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 31187
    },
    {
      "epoch": 12.460247702756693,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 31188
    },
    {
      "epoch": 12.460647223332002,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 31189
    },
    {
      "epoch": 12.46104674390731,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 31190
    },
    {
      "epoch": 12.461446264482621,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 31191
    },
    {
      "epoch": 12.46184578505793,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 31192
    },
    {
      "epoch": 12.46224530563324,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 31193
    },
    {
      "epoch": 12.46264482620855,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 31194
    },
    {
      "epoch": 12.463044346783859,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 31195
    },
    {
      "epoch": 12.46344386735917,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 31196
    },
    {
      "epoch": 12.463843387934478,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 31197
    },
    {
      "epoch": 12.464242908509789,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 31198
    },
    {
      "epoch": 12.464642429085098,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 31199
    },
    {
      "epoch": 12.465041949660408,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 31200
    },
    {
      "epoch": 12.465441470235717,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 31201
    },
    {
      "epoch": 12.465840990811026,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 31202
    },
    {
      "epoch": 12.466240511386337,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 31203
    },
    {
      "epoch": 12.466640031961646,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 31204
    },
    {
      "epoch": 12.467039552536956,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 31205
    },
    {
      "epoch": 12.467439073112265,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 31206
    },
    {
      "epoch": 12.467838593687574,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 31207
    },
    {
      "epoch": 12.468238114262885,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 31208
    },
    {
      "epoch": 12.468637634838194,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 31209
    },
    {
      "epoch": 12.469037155413504,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 31210
    },
    {
      "epoch": 12.469436675988813,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 31211
    },
    {
      "epoch": 12.469836196564122,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 31212
    },
    {
      "epoch": 12.470235717139433,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 31213
    },
    {
      "epoch": 12.470635237714742,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 31214
    },
    {
      "epoch": 12.471034758290052,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 31215
    },
    {
      "epoch": 12.471434278865361,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 31216
    },
    {
      "epoch": 12.471833799440672,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 31217
    },
    {
      "epoch": 12.472233320015981,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 31218
    },
    {
      "epoch": 12.47263284059129,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 31219
    },
    {
      "epoch": 12.4730323611666,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 31220
    },
    {
      "epoch": 12.47343188174191,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 31221
    },
    {
      "epoch": 12.47383140231722,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 31222
    },
    {
      "epoch": 12.474230922892529,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 31223
    },
    {
      "epoch": 12.474630443467838,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 31224
    },
    {
      "epoch": 12.475029964043149,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 31225
    },
    {
      "epoch": 12.475429484618457,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 31226
    },
    {
      "epoch": 12.475829005193768,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 31227
    },
    {
      "epoch": 12.476228525769077,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 31228
    },
    {
      "epoch": 12.476628046344386,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 31229
    },
    {
      "epoch": 12.477027566919697,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 31230
    },
    {
      "epoch": 12.477427087495006,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 31231
    },
    {
      "epoch": 12.477826608070316,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 31232
    },
    {
      "epoch": 12.478226128645625,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 31233
    },
    {
      "epoch": 12.478625649220934,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 31234
    },
    {
      "epoch": 12.479025169796245,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 31235
    },
    {
      "epoch": 12.479424690371554,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 31236
    },
    {
      "epoch": 12.479824210946864,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 31237
    },
    {
      "epoch": 12.480223731522173,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 31238
    },
    {
      "epoch": 12.480623252097484,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 31239
    },
    {
      "epoch": 12.481022772672793,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 31240
    },
    {
      "epoch": 12.481422293248102,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 31241
    },
    {
      "epoch": 12.481821813823412,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 31242
    },
    {
      "epoch": 12.482221334398721,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 31243
    },
    {
      "epoch": 12.482620854974032,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 31244
    },
    {
      "epoch": 12.48302037554934,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 31245
    },
    {
      "epoch": 12.48341989612465,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 31246
    },
    {
      "epoch": 12.48381941669996,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 31247
    },
    {
      "epoch": 12.48421893727527,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 31248
    },
    {
      "epoch": 12.48461845785058,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 31249
    },
    {
      "epoch": 12.485017978425889,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 31250
    },
    {
      "epoch": 12.485417499001198,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 31251
    },
    {
      "epoch": 12.485817019576508,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 31252
    },
    {
      "epoch": 12.486216540151817,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 31253
    },
    {
      "epoch": 12.486616060727128,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 31254
    },
    {
      "epoch": 12.487015581302437,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 31255
    },
    {
      "epoch": 12.487415101877748,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 31256
    },
    {
      "epoch": 12.487814622453056,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 31257
    },
    {
      "epoch": 12.488214143028365,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 31258
    },
    {
      "epoch": 12.488613663603676,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 31259
    },
    {
      "epoch": 12.489013184178985,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 31260
    },
    {
      "epoch": 12.489412704754296,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 31261
    },
    {
      "epoch": 12.489812225329604,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 31262
    },
    {
      "epoch": 12.490211745904913,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 31263
    },
    {
      "epoch": 12.490611266480224,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 31264
    },
    {
      "epoch": 12.491010787055533,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 31265
    },
    {
      "epoch": 12.491410307630844,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 31266
    },
    {
      "epoch": 12.491809828206152,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 31267
    },
    {
      "epoch": 12.492209348781461,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 31268
    },
    {
      "epoch": 12.492608869356772,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 31269
    },
    {
      "epoch": 12.493008389932081,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 31270
    },
    {
      "epoch": 12.493407910507392,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 31271
    },
    {
      "epoch": 12.4938074310827,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 31272
    },
    {
      "epoch": 12.494206951658011,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 31273
    },
    {
      "epoch": 12.49460647223332,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 31274
    },
    {
      "epoch": 12.495005992808629,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 31275
    },
    {
      "epoch": 12.49540551338394,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 31276
    },
    {
      "epoch": 12.495805033959249,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 31277
    },
    {
      "epoch": 12.49620455453456,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 31278
    },
    {
      "epoch": 12.496604075109868,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 31279
    },
    {
      "epoch": 12.497003595685177,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 31280
    },
    {
      "epoch": 12.497403116260488,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 31281
    },
    {
      "epoch": 12.497802636835797,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 31282
    },
    {
      "epoch": 12.498202157411107,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 31283
    },
    {
      "epoch": 12.498601677986416,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 31284
    },
    {
      "epoch": 12.499001198561725,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 31285
    },
    {
      "epoch": 12.499400719137036,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 31286
    },
    {
      "epoch": 12.499800239712345,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 31287
    },
    {
      "epoch": 12.500199760287655,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 31288
    },
    {
      "epoch": 12.500599280862964,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 31289
    },
    {
      "epoch": 12.500998801438275,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 31290
    },
    {
      "epoch": 12.501398322013584,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 31291
    },
    {
      "epoch": 12.501797842588893,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 31292
    },
    {
      "epoch": 12.502197363164203,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 31293
    },
    {
      "epoch": 12.502596883739512,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 31294
    },
    {
      "epoch": 12.502996404314823,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 31295
    },
    {
      "epoch": 12.503395924890132,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 31296
    },
    {
      "epoch": 12.50379544546544,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 31297
    },
    {
      "epoch": 12.504194966040751,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 31298
    },
    {
      "epoch": 12.50459448661606,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 31299
    },
    {
      "epoch": 12.504994007191371,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 31300
    },
    {
      "epoch": 12.50539352776668,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 31301
    },
    {
      "epoch": 12.505793048341989,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 31302
    },
    {
      "epoch": 12.5061925689173,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 31303
    },
    {
      "epoch": 12.506592089492608,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 31304
    },
    {
      "epoch": 12.506991610067919,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 31305
    },
    {
      "epoch": 12.507391130643228,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 31306
    },
    {
      "epoch": 12.507790651218539,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 31307
    },
    {
      "epoch": 12.508190171793848,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 31308
    },
    {
      "epoch": 12.508589692369156,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 31309
    },
    {
      "epoch": 12.508989212944467,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 31310
    },
    {
      "epoch": 12.509388733519776,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 31311
    },
    {
      "epoch": 12.509788254095087,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 31312
    },
    {
      "epoch": 12.510187774670396,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 31313
    },
    {
      "epoch": 12.510587295245704,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 31314
    },
    {
      "epoch": 12.510986815821015,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 31315
    },
    {
      "epoch": 12.511386336396324,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 31316
    },
    {
      "epoch": 12.511785856971635,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 31317
    },
    {
      "epoch": 12.512185377546944,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 31318
    },
    {
      "epoch": 12.512584898122252,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 31319
    },
    {
      "epoch": 12.512984418697563,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 31320
    },
    {
      "epoch": 12.513383939272872,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 31321
    },
    {
      "epoch": 12.513783459848183,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 31322
    },
    {
      "epoch": 12.514182980423492,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 31323
    },
    {
      "epoch": 12.514582500998802,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 31324
    },
    {
      "epoch": 12.514982021574111,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 31325
    },
    {
      "epoch": 12.51538154214942,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 31326
    },
    {
      "epoch": 12.51578106272473,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 31327
    },
    {
      "epoch": 12.51618058330004,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 31328
    },
    {
      "epoch": 12.51658010387535,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 31329
    },
    {
      "epoch": 12.51697962445066,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 31330
    },
    {
      "epoch": 12.517379145025968,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 31331
    },
    {
      "epoch": 12.517778665601279,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 31332
    },
    {
      "epoch": 12.518178186176588,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 31333
    },
    {
      "epoch": 12.518577706751898,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 31334
    },
    {
      "epoch": 12.518977227327207,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 31335
    },
    {
      "epoch": 12.519376747902516,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 31336
    },
    {
      "epoch": 12.519776268477827,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 31337
    },
    {
      "epoch": 12.520175789053136,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 31338
    },
    {
      "epoch": 12.520575309628446,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 31339
    },
    {
      "epoch": 12.520974830203755,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 31340
    },
    {
      "epoch": 12.521374350779066,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 31341
    },
    {
      "epoch": 12.521773871354375,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 31342
    },
    {
      "epoch": 12.522173391929684,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 31343
    },
    {
      "epoch": 12.522572912504994,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 31344
    },
    {
      "epoch": 12.522972433080303,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 31345
    },
    {
      "epoch": 12.523371953655614,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 31346
    },
    {
      "epoch": 12.523771474230923,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 31347
    },
    {
      "epoch": 12.524170994806232,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 31348
    },
    {
      "epoch": 12.524570515381543,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 31349
    },
    {
      "epoch": 12.524970035956851,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 31350
    },
    {
      "epoch": 12.525369556532162,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 31351
    },
    {
      "epoch": 12.525769077107471,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 31352
    },
    {
      "epoch": 12.52616859768278,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 31353
    },
    {
      "epoch": 12.52656811825809,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 31354
    },
    {
      "epoch": 12.5269676388334,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 31355
    },
    {
      "epoch": 12.52736715940871,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 31356
    },
    {
      "epoch": 12.527766679984019,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 31357
    },
    {
      "epoch": 12.52816620055933,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 31358
    },
    {
      "epoch": 12.528565721134639,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 31359
    },
    {
      "epoch": 12.528965241709948,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 31360
    },
    {
      "epoch": 12.529364762285258,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 31361
    },
    {
      "epoch": 12.529764282860567,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 31362
    },
    {
      "epoch": 12.530163803435878,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 31363
    },
    {
      "epoch": 12.530563324011187,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 31364
    },
    {
      "epoch": 12.530962844586496,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 31365
    },
    {
      "epoch": 12.531362365161806,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 31366
    },
    {
      "epoch": 12.531761885737115,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 31367
    },
    {
      "epoch": 12.532161406312426,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 31368
    },
    {
      "epoch": 12.532560926887735,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 31369
    },
    {
      "epoch": 12.532960447463044,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 31370
    },
    {
      "epoch": 12.533359968038354,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 31371
    },
    {
      "epoch": 12.533759488613663,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 31372
    },
    {
      "epoch": 12.534159009188974,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 31373
    },
    {
      "epoch": 12.534558529764283,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 31374
    },
    {
      "epoch": 12.534958050339593,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 31375
    },
    {
      "epoch": 12.535357570914902,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 31376
    },
    {
      "epoch": 12.535757091490211,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 31377
    },
    {
      "epoch": 12.536156612065522,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 31378
    },
    {
      "epoch": 12.53655613264083,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 31379
    },
    {
      "epoch": 12.53695565321614,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 31380
    },
    {
      "epoch": 12.53735517379145,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 31381
    },
    {
      "epoch": 12.53775469436676,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 31382
    },
    {
      "epoch": 12.53815421494207,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 31383
    },
    {
      "epoch": 12.538553735517379,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 31384
    },
    {
      "epoch": 12.53895325609269,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 31385
    },
    {
      "epoch": 12.539352776667998,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 31386
    },
    {
      "epoch": 12.539752297243307,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 31387
    },
    {
      "epoch": 12.540151817818618,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 31388
    },
    {
      "epoch": 12.540551338393927,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 31389
    },
    {
      "epoch": 12.540950858969238,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 31390
    },
    {
      "epoch": 12.541350379544546,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 31391
    },
    {
      "epoch": 12.541749900119855,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 31392
    },
    {
      "epoch": 12.542149420695166,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 31393
    },
    {
      "epoch": 12.542548941270475,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 31394
    },
    {
      "epoch": 12.542948461845786,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 31395
    },
    {
      "epoch": 12.543347982421094,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 31396
    },
    {
      "epoch": 12.543747502996403,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 31397
    },
    {
      "epoch": 12.544147023571714,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 31398
    },
    {
      "epoch": 12.544546544147023,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 31399
    },
    {
      "epoch": 12.544946064722334,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 31400
    },
    {
      "epoch": 12.545345585297643,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 31401
    },
    {
      "epoch": 12.545745105872953,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 31402
    },
    {
      "epoch": 12.546144626448262,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 31403
    },
    {
      "epoch": 12.546544147023571,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 31404
    },
    {
      "epoch": 12.546943667598882,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 31405
    },
    {
      "epoch": 12.54734318817419,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 31406
    },
    {
      "epoch": 12.547742708749501,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 31407
    },
    {
      "epoch": 12.54814222932481,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 31408
    },
    {
      "epoch": 12.548541749900119,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 31409
    },
    {
      "epoch": 12.54894127047543,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 31410
    },
    {
      "epoch": 12.549340791050739,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 31411
    },
    {
      "epoch": 12.54974031162605,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 31412
    },
    {
      "epoch": 12.550139832201358,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 31413
    },
    {
      "epoch": 12.550539352776667,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 31414
    },
    {
      "epoch": 12.550938873351978,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 31415
    },
    {
      "epoch": 12.551338393927287,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 31416
    },
    {
      "epoch": 12.551737914502597,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 31417
    },
    {
      "epoch": 12.552137435077906,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 31418
    },
    {
      "epoch": 12.552536955653217,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 31419
    },
    {
      "epoch": 12.552936476228526,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 31420
    },
    {
      "epoch": 12.553335996803835,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 31421
    },
    {
      "epoch": 12.553735517379145,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 31422
    },
    {
      "epoch": 12.554135037954454,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 31423
    },
    {
      "epoch": 12.554534558529765,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 31424
    },
    {
      "epoch": 12.554934079105074,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 31425
    },
    {
      "epoch": 12.555333599680383,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 31426
    },
    {
      "epoch": 12.555733120255693,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 31427
    },
    {
      "epoch": 12.556132640831002,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 31428
    },
    {
      "epoch": 12.556532161406313,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 31429
    },
    {
      "epoch": 12.556931681981622,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 31430
    },
    {
      "epoch": 12.55733120255693,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 31431
    },
    {
      "epoch": 12.557730723132241,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 31432
    },
    {
      "epoch": 12.55813024370755,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 31433
    },
    {
      "epoch": 12.558529764282861,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 31434
    },
    {
      "epoch": 12.55892928485817,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 31435
    },
    {
      "epoch": 12.55932880543348,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 31436
    },
    {
      "epoch": 12.55972832600879,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 31437
    },
    {
      "epoch": 12.560127846584098,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 31438
    },
    {
      "epoch": 12.560527367159409,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 31439
    },
    {
      "epoch": 12.560926887734718,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 31440
    },
    {
      "epoch": 12.561326408310029,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 31441
    },
    {
      "epoch": 12.561725928885338,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 31442
    },
    {
      "epoch": 12.562125449460646,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 31443
    },
    {
      "epoch": 12.562524970035957,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 31444
    },
    {
      "epoch": 12.562924490611266,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 31445
    },
    {
      "epoch": 12.563324011186577,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 31446
    },
    {
      "epoch": 12.563723531761886,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 31447
    },
    {
      "epoch": 12.564123052337195,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 31448
    },
    {
      "epoch": 12.564522572912505,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 31449
    },
    {
      "epoch": 12.564922093487814,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 31450
    },
    {
      "epoch": 12.565321614063125,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 31451
    },
    {
      "epoch": 12.565721134638434,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 31452
    },
    {
      "epoch": 12.566120655213744,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 31453
    },
    {
      "epoch": 12.566520175789053,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 31454
    },
    {
      "epoch": 12.566919696364362,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 31455
    },
    {
      "epoch": 12.567319216939673,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 31456
    },
    {
      "epoch": 12.567718737514982,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 31457
    },
    {
      "epoch": 12.568118258090292,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 31458
    },
    {
      "epoch": 12.568517778665601,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 31459
    },
    {
      "epoch": 12.56891729924091,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 31460
    },
    {
      "epoch": 12.56931681981622,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 31461
    },
    {
      "epoch": 12.56971634039153,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 31462
    },
    {
      "epoch": 12.57011586096684,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 31463
    },
    {
      "epoch": 12.57051538154215,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 31464
    },
    {
      "epoch": 12.570914902117458,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 31465
    },
    {
      "epoch": 12.571314422692769,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 31466
    },
    {
      "epoch": 12.571713943268078,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 31467
    },
    {
      "epoch": 12.572113463843388,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 31468
    },
    {
      "epoch": 12.572512984418697,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 31469
    },
    {
      "epoch": 12.572912504994008,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 31470
    },
    {
      "epoch": 12.573312025569317,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 31471
    },
    {
      "epoch": 12.573711546144626,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 31472
    },
    {
      "epoch": 12.574111066719937,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 31473
    },
    {
      "epoch": 12.574510587295245,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 31474
    },
    {
      "epoch": 12.574910107870556,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 31475
    },
    {
      "epoch": 12.575309628445865,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 31476
    },
    {
      "epoch": 12.575709149021174,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 31477
    },
    {
      "epoch": 12.576108669596485,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 31478
    },
    {
      "epoch": 12.576508190171793,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 31479
    },
    {
      "epoch": 12.576907710747104,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 31480
    },
    {
      "epoch": 12.577307231322413,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 31481
    },
    {
      "epoch": 12.577706751897722,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 31482
    },
    {
      "epoch": 12.578106272473033,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 31483
    },
    {
      "epoch": 12.578505793048341,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 31484
    },
    {
      "epoch": 12.578905313623652,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 31485
    },
    {
      "epoch": 12.579304834198961,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 31486
    },
    {
      "epoch": 12.579704354774272,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 31487
    },
    {
      "epoch": 12.58010387534958,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 31488
    },
    {
      "epoch": 12.58050339592489,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 31489
    },
    {
      "epoch": 12.5809029165002,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 31490
    },
    {
      "epoch": 12.581302437075509,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 31491
    },
    {
      "epoch": 12.58170195765082,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 31492
    },
    {
      "epoch": 12.582101478226129,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 31493
    },
    {
      "epoch": 12.582500998801438,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 31494
    },
    {
      "epoch": 12.582900519376748,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 31495
    },
    {
      "epoch": 12.583300039952057,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 31496
    },
    {
      "epoch": 12.583699560527368,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 31497
    },
    {
      "epoch": 12.584099081102677,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 31498
    },
    {
      "epoch": 12.584498601677986,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 31499
    },
    {
      "epoch": 12.584898122253296,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 31500
    },
    {
      "epoch": 12.585297642828605,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 31501
    },
    {
      "epoch": 12.585697163403916,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 31502
    },
    {
      "epoch": 12.586096683979225,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 31503
    },
    {
      "epoch": 12.586496204554535,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 31504
    },
    {
      "epoch": 12.586895725129844,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 31505
    },
    {
      "epoch": 12.587295245705153,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 31506
    },
    {
      "epoch": 12.587694766280464,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 31507
    },
    {
      "epoch": 12.588094286855773,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 31508
    },
    {
      "epoch": 12.588493807431083,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 31509
    },
    {
      "epoch": 12.588893328006392,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 31510
    },
    {
      "epoch": 12.589292848581701,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 31511
    },
    {
      "epoch": 12.589692369157012,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 31512
    },
    {
      "epoch": 12.59009188973232,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 31513
    },
    {
      "epoch": 12.590491410307632,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 31514
    },
    {
      "epoch": 12.59089093088294,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 31515
    },
    {
      "epoch": 12.59129045145825,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 31516
    },
    {
      "epoch": 12.59168997203356,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 31517
    },
    {
      "epoch": 12.592089492608869,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 31518
    },
    {
      "epoch": 12.59248901318418,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 31519
    },
    {
      "epoch": 12.592888533759488,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 31520
    },
    {
      "epoch": 12.5932880543348,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 31521
    },
    {
      "epoch": 12.593687574910108,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 31522
    },
    {
      "epoch": 12.594087095485417,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 31523
    },
    {
      "epoch": 12.594486616060728,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 31524
    },
    {
      "epoch": 12.594886136636037,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 31525
    },
    {
      "epoch": 12.595285657211347,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 31526
    },
    {
      "epoch": 12.595685177786656,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 31527
    },
    {
      "epoch": 12.596084698361965,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 31528
    },
    {
      "epoch": 12.596484218937276,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 31529
    },
    {
      "epoch": 12.596883739512585,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 31530
    },
    {
      "epoch": 12.597283260087895,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 31531
    },
    {
      "epoch": 12.597682780663204,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 31532
    },
    {
      "epoch": 12.598082301238513,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 31533
    },
    {
      "epoch": 12.598481821813824,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 31534
    },
    {
      "epoch": 12.598881342389133,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 31535
    },
    {
      "epoch": 12.599280862964443,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 31536
    },
    {
      "epoch": 12.599680383539752,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 31537
    },
    {
      "epoch": 12.600079904115063,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 31538
    },
    {
      "epoch": 12.600479424690372,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 31539
    },
    {
      "epoch": 12.60087894526568,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 31540
    },
    {
      "epoch": 12.601278465840991,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 31541
    },
    {
      "epoch": 12.6016779864163,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 31542
    },
    {
      "epoch": 12.602077506991611,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 31543
    },
    {
      "epoch": 12.60247702756692,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 31544
    },
    {
      "epoch": 12.602876548142229,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 31545
    },
    {
      "epoch": 12.60327606871754,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 31546
    },
    {
      "epoch": 12.603675589292848,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 31547
    },
    {
      "epoch": 12.604075109868159,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 31548
    },
    {
      "epoch": 12.604474630443468,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 31549
    },
    {
      "epoch": 12.604874151018777,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 31550
    },
    {
      "epoch": 12.605273671594087,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 31551
    },
    {
      "epoch": 12.605673192169396,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 31552
    },
    {
      "epoch": 12.606072712744707,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 31553
    },
    {
      "epoch": 12.606472233320016,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 31554
    },
    {
      "epoch": 12.606871753895327,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 31555
    },
    {
      "epoch": 12.607271274470635,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 31556
    },
    {
      "epoch": 12.607670795045944,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 31557
    },
    {
      "epoch": 12.608070315621255,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 31558
    },
    {
      "epoch": 12.608469836196564,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 31559
    },
    {
      "epoch": 12.608869356771873,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 31560
    },
    {
      "epoch": 12.609268877347183,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 31561
    },
    {
      "epoch": 12.609668397922492,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 31562
    },
    {
      "epoch": 12.610067918497803,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 31563
    },
    {
      "epoch": 12.610467439073112,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 31564
    },
    {
      "epoch": 12.610866959648423,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 31565
    },
    {
      "epoch": 12.611266480223732,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 31566
    },
    {
      "epoch": 12.61166600079904,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 31567
    },
    {
      "epoch": 12.612065521374351,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 31568
    },
    {
      "epoch": 12.61246504194966,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 31569
    },
    {
      "epoch": 12.61286456252497,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 31570
    },
    {
      "epoch": 12.61326408310028,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 31571
    },
    {
      "epoch": 12.613663603675588,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 31572
    },
    {
      "epoch": 12.6140631242509,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 31573
    },
    {
      "epoch": 12.614462644826208,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 31574
    },
    {
      "epoch": 12.614862165401519,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 31575
    },
    {
      "epoch": 12.615261685976828,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 31576
    },
    {
      "epoch": 12.615661206552137,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 31577
    },
    {
      "epoch": 12.616060727127447,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 31578
    },
    {
      "epoch": 12.616460247702756,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 31579
    },
    {
      "epoch": 12.616859768278067,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 31580
    },
    {
      "epoch": 12.617259288853376,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 31581
    },
    {
      "epoch": 12.617658809428686,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 31582
    },
    {
      "epoch": 12.618058330003995,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 31583
    },
    {
      "epoch": 12.618457850579304,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 31584
    },
    {
      "epoch": 12.618857371154615,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 31585
    },
    {
      "epoch": 12.619256891729924,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 31586
    },
    {
      "epoch": 12.619656412305234,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 31587
    },
    {
      "epoch": 12.620055932880543,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 31588
    },
    {
      "epoch": 12.620455453455852,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 31589
    },
    {
      "epoch": 12.620854974031163,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 31590
    },
    {
      "epoch": 12.621254494606472,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 31591
    },
    {
      "epoch": 12.621654015181782,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 31592
    },
    {
      "epoch": 12.622053535757091,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 31593
    },
    {
      "epoch": 12.6224530563324,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 31594
    },
    {
      "epoch": 12.622852576907711,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 31595
    },
    {
      "epoch": 12.62325209748302,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 31596
    },
    {
      "epoch": 12.62365161805833,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 31597
    },
    {
      "epoch": 12.62405113863364,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 31598
    },
    {
      "epoch": 12.62445065920895,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 31599
    },
    {
      "epoch": 12.624850179784259,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 31600
    },
    {
      "epoch": 12.625249700359568,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 31601
    },
    {
      "epoch": 12.625649220934879,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 31602
    },
    {
      "epoch": 12.626048741510187,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 31603
    },
    {
      "epoch": 12.626448262085498,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 31604
    },
    {
      "epoch": 12.626847782660807,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 31605
    },
    {
      "epoch": 12.627247303236116,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 31606
    },
    {
      "epoch": 12.627646823811427,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 31607
    },
    {
      "epoch": 12.628046344386735,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 31608
    },
    {
      "epoch": 12.628445864962046,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 31609
    },
    {
      "epoch": 12.628845385537355,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 31610
    },
    {
      "epoch": 12.629244906112664,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 31611
    },
    {
      "epoch": 12.629644426687975,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 31612
    },
    {
      "epoch": 12.630043947263283,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 31613
    },
    {
      "epoch": 12.630443467838594,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 31614
    },
    {
      "epoch": 12.630842988413903,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 31615
    },
    {
      "epoch": 12.631242508989214,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 31616
    },
    {
      "epoch": 12.631642029564523,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 31617
    },
    {
      "epoch": 12.632041550139832,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 31618
    },
    {
      "epoch": 12.632441070715142,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 31619
    },
    {
      "epoch": 12.632840591290451,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 31620
    },
    {
      "epoch": 12.633240111865762,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 31621
    },
    {
      "epoch": 12.63363963244107,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 31622
    },
    {
      "epoch": 12.63403915301638,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 31623
    },
    {
      "epoch": 12.63443867359169,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 31624
    },
    {
      "epoch": 12.634838194167,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 31625
    },
    {
      "epoch": 12.63523771474231,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 31626
    },
    {
      "epoch": 12.635637235317619,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 31627
    },
    {
      "epoch": 12.636036755892928,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 31628
    },
    {
      "epoch": 12.636436276468238,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 31629
    },
    {
      "epoch": 12.636835797043547,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 31630
    },
    {
      "epoch": 12.637235317618858,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 31631
    },
    {
      "epoch": 12.637634838194167,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 31632
    },
    {
      "epoch": 12.638034358769477,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 31633
    },
    {
      "epoch": 12.638433879344786,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 31634
    },
    {
      "epoch": 12.638833399920095,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 31635
    },
    {
      "epoch": 12.639232920495406,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 31636
    },
    {
      "epoch": 12.639632441070715,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 31637
    },
    {
      "epoch": 12.640031961646025,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 31638
    },
    {
      "epoch": 12.640431482221334,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 31639
    },
    {
      "epoch": 12.640831002796643,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 31640
    },
    {
      "epoch": 12.641230523371954,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 31641
    },
    {
      "epoch": 12.641630043947263,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 31642
    },
    {
      "epoch": 12.642029564522574,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 31643
    },
    {
      "epoch": 12.642429085097882,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 31644
    },
    {
      "epoch": 12.642828605673191,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 31645
    },
    {
      "epoch": 12.643228126248502,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 31646
    },
    {
      "epoch": 12.643627646823811,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 31647
    },
    {
      "epoch": 12.644027167399122,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 31648
    },
    {
      "epoch": 12.64442668797443,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 31649
    },
    {
      "epoch": 12.644826208549741,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 31650
    },
    {
      "epoch": 12.64522572912505,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 31651
    },
    {
      "epoch": 12.645625249700359,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 31652
    },
    {
      "epoch": 12.64602477027567,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 31653
    },
    {
      "epoch": 12.646424290850979,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 31654
    },
    {
      "epoch": 12.64682381142629,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 31655
    },
    {
      "epoch": 12.647223332001598,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 31656
    },
    {
      "epoch": 12.647622852576907,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 31657
    },
    {
      "epoch": 12.648022373152218,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 31658
    },
    {
      "epoch": 12.648421893727527,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 31659
    },
    {
      "epoch": 12.648821414302837,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 31660
    },
    {
      "epoch": 12.649220934878146,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 31661
    },
    {
      "epoch": 12.649620455453455,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 31662
    },
    {
      "epoch": 12.650019976028766,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 31663
    },
    {
      "epoch": 12.650419496604075,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 31664
    },
    {
      "epoch": 12.650819017179385,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 31665
    },
    {
      "epoch": 12.651218537754694,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 31666
    },
    {
      "epoch": 12.651618058330005,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 31667
    },
    {
      "epoch": 12.652017578905314,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 31668
    },
    {
      "epoch": 12.652417099480623,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 31669
    },
    {
      "epoch": 12.652816620055933,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0112,
      "step": 31670
    },
    {
      "epoch": 12.653216140631242,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 31671
    },
    {
      "epoch": 12.653615661206553,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 31672
    },
    {
      "epoch": 12.654015181781862,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 31673
    },
    {
      "epoch": 12.65441470235717,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 31674
    },
    {
      "epoch": 12.654814222932481,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 31675
    },
    {
      "epoch": 12.65521374350779,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 31676
    },
    {
      "epoch": 12.655613264083101,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 31677
    },
    {
      "epoch": 12.65601278465841,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 31678
    },
    {
      "epoch": 12.656412305233719,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 31679
    },
    {
      "epoch": 12.65681182580903,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 31680
    },
    {
      "epoch": 12.657211346384338,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 31681
    },
    {
      "epoch": 12.657610866959649,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 31682
    },
    {
      "epoch": 12.658010387534958,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 31683
    },
    {
      "epoch": 12.658409908110269,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 31684
    },
    {
      "epoch": 12.658809428685577,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 31685
    },
    {
      "epoch": 12.659208949260886,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 31686
    },
    {
      "epoch": 12.659608469836197,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 31687
    },
    {
      "epoch": 12.660007990411506,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 31688
    },
    {
      "epoch": 12.660407510986817,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 31689
    },
    {
      "epoch": 12.660807031562126,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 31690
    },
    {
      "epoch": 12.661206552137434,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 31691
    },
    {
      "epoch": 12.661606072712745,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 31692
    },
    {
      "epoch": 12.662005593288054,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 31693
    },
    {
      "epoch": 12.662405113863365,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 31694
    },
    {
      "epoch": 12.662804634438674,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 31695
    },
    {
      "epoch": 12.663204155013982,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 31696
    },
    {
      "epoch": 12.663603675589293,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 31697
    },
    {
      "epoch": 12.664003196164602,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 31698
    },
    {
      "epoch": 12.664402716739913,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 31699
    },
    {
      "epoch": 12.664802237315222,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 31700
    },
    {
      "epoch": 12.665201757890532,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 31701
    },
    {
      "epoch": 12.665601278465841,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 31702
    },
    {
      "epoch": 12.66600079904115,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 31703
    },
    {
      "epoch": 12.66640031961646,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 31704
    },
    {
      "epoch": 12.66679984019177,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 31705
    },
    {
      "epoch": 12.66719936076708,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 31706
    },
    {
      "epoch": 12.66759888134239,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 31707
    },
    {
      "epoch": 12.667998401917698,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 31708
    },
    {
      "epoch": 12.668397922493009,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 31709
    },
    {
      "epoch": 12.668797443068318,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 31710
    },
    {
      "epoch": 12.669196963643628,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 31711
    },
    {
      "epoch": 12.669596484218937,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 31712
    },
    {
      "epoch": 12.669996004794246,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 31713
    },
    {
      "epoch": 12.670395525369557,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 31714
    },
    {
      "epoch": 12.670795045944866,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 31715
    },
    {
      "epoch": 12.671194566520176,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 31716
    },
    {
      "epoch": 12.671594087095485,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 31717
    },
    {
      "epoch": 12.671993607670796,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 31718
    },
    {
      "epoch": 12.672393128246105,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 31719
    },
    {
      "epoch": 12.672792648821414,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 31720
    },
    {
      "epoch": 12.673192169396724,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 31721
    },
    {
      "epoch": 12.673591689972033,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 31722
    },
    {
      "epoch": 12.673991210547344,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 31723
    },
    {
      "epoch": 12.674390731122653,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 31724
    },
    {
      "epoch": 12.674790251697962,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 31725
    },
    {
      "epoch": 12.675189772273272,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 31726
    },
    {
      "epoch": 12.675589292848581,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 31727
    },
    {
      "epoch": 12.675988813423892,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 31728
    },
    {
      "epoch": 12.676388333999201,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 31729
    },
    {
      "epoch": 12.67678785457451,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 31730
    },
    {
      "epoch": 12.67718737514982,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 31731
    },
    {
      "epoch": 12.67758689572513,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 31732
    },
    {
      "epoch": 12.67798641630044,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 31733
    },
    {
      "epoch": 12.678385936875749,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 31734
    },
    {
      "epoch": 12.67878545745106,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 31735
    },
    {
      "epoch": 12.679184978026369,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 31736
    },
    {
      "epoch": 12.679584498601677,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 31737
    },
    {
      "epoch": 12.679984019176988,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 31738
    },
    {
      "epoch": 12.680383539752297,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 31739
    },
    {
      "epoch": 12.680783060327606,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 31740
    },
    {
      "epoch": 12.681182580902917,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 31741
    },
    {
      "epoch": 12.681582101478226,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 31742
    },
    {
      "epoch": 12.681981622053536,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 31743
    },
    {
      "epoch": 12.682381142628845,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 31744
    },
    {
      "epoch": 12.682780663204156,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 31745
    },
    {
      "epoch": 12.683180183779465,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 31746
    },
    {
      "epoch": 12.683579704354774,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 31747
    },
    {
      "epoch": 12.683979224930084,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 31748
    },
    {
      "epoch": 12.684378745505393,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 31749
    },
    {
      "epoch": 12.684778266080704,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 31750
    },
    {
      "epoch": 12.685177786656013,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 31751
    },
    {
      "epoch": 12.685577307231322,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 31752
    },
    {
      "epoch": 12.685976827806632,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 31753
    },
    {
      "epoch": 12.686376348381941,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0439,
      "step": 31754
    },
    {
      "epoch": 12.686775868957252,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 31755
    },
    {
      "epoch": 12.68717538953256,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 31756
    },
    {
      "epoch": 12.68757491010787,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 31757
    },
    {
      "epoch": 12.68797443068318,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 31758
    },
    {
      "epoch": 12.68837395125849,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 31759
    },
    {
      "epoch": 12.6887734718338,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 31760
    },
    {
      "epoch": 12.689172992409109,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 31761
    },
    {
      "epoch": 12.68957251298442,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 31762
    },
    {
      "epoch": 12.689972033559728,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 31763
    },
    {
      "epoch": 12.690371554135037,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 31764
    },
    {
      "epoch": 12.690771074710348,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 31765
    },
    {
      "epoch": 12.691170595285657,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 31766
    },
    {
      "epoch": 12.691570115860968,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 31767
    },
    {
      "epoch": 12.691969636436276,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 31768
    },
    {
      "epoch": 12.692369157011585,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 31769
    },
    {
      "epoch": 12.692768677586896,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 31770
    },
    {
      "epoch": 12.693168198162205,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 31771
    },
    {
      "epoch": 12.693567718737516,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 31772
    },
    {
      "epoch": 12.693967239312824,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 31773
    },
    {
      "epoch": 12.694366759888133,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 31774
    },
    {
      "epoch": 12.694766280463444,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 31775
    },
    {
      "epoch": 12.695165801038753,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 31776
    },
    {
      "epoch": 12.695565321614064,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 31777
    },
    {
      "epoch": 12.695964842189372,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 31778
    },
    {
      "epoch": 12.696364362764683,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 31779
    },
    {
      "epoch": 12.696763883339992,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 31780
    },
    {
      "epoch": 12.697163403915301,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 31781
    },
    {
      "epoch": 12.697562924490612,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 31782
    },
    {
      "epoch": 12.69796244506592,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 31783
    },
    {
      "epoch": 12.698361965641231,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 31784
    },
    {
      "epoch": 12.69876148621654,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 31785
    },
    {
      "epoch": 12.699161006791849,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 31786
    },
    {
      "epoch": 12.69956052736716,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 31787
    },
    {
      "epoch": 12.699960047942469,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 31788
    },
    {
      "epoch": 12.70035956851778,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 31789
    },
    {
      "epoch": 12.700759089093088,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 31790
    },
    {
      "epoch": 12.701158609668397,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 31791
    },
    {
      "epoch": 12.701558130243708,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 31792
    },
    {
      "epoch": 12.701957650819017,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 31793
    },
    {
      "epoch": 12.702357171394327,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 31794
    },
    {
      "epoch": 12.702756691969636,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 31795
    },
    {
      "epoch": 12.703156212544947,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 31796
    },
    {
      "epoch": 12.703555733120256,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 31797
    },
    {
      "epoch": 12.703955253695565,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 31798
    },
    {
      "epoch": 12.704354774270875,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 31799
    },
    {
      "epoch": 12.704754294846184,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 31800
    },
    {
      "epoch": 12.705153815421495,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 31801
    },
    {
      "epoch": 12.705553335996804,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 31802
    },
    {
      "epoch": 12.705952856572113,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0563,
      "step": 31803
    },
    {
      "epoch": 12.706352377147423,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 31804
    },
    {
      "epoch": 12.706751897722732,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 31805
    },
    {
      "epoch": 12.707151418298043,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 31806
    },
    {
      "epoch": 12.707550938873352,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 31807
    },
    {
      "epoch": 12.70795045944866,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 31808
    },
    {
      "epoch": 12.708349980023971,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 31809
    },
    {
      "epoch": 12.70874950059928,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 31810
    },
    {
      "epoch": 12.709149021174591,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 31811
    },
    {
      "epoch": 12.7095485417499,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 31812
    },
    {
      "epoch": 12.70994806232521,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 31813
    },
    {
      "epoch": 12.71034758290052,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 31814
    },
    {
      "epoch": 12.710747103475828,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 31815
    },
    {
      "epoch": 12.711146624051139,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 31816
    },
    {
      "epoch": 12.711546144626448,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 31817
    },
    {
      "epoch": 12.711945665201759,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 31818
    },
    {
      "epoch": 12.712345185777068,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 31819
    },
    {
      "epoch": 12.712744706352376,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 31820
    },
    {
      "epoch": 12.713144226927687,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 31821
    },
    {
      "epoch": 12.713543747502996,
      "grad_norm": 0.171875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 31822
    },
    {
      "epoch": 12.713943268078307,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 31823
    },
    {
      "epoch": 12.714342788653616,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 31824
    },
    {
      "epoch": 12.714742309228924,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 31825
    },
    {
      "epoch": 12.715141829804235,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 31826
    },
    {
      "epoch": 12.715541350379544,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 31827
    },
    {
      "epoch": 12.715940870954855,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 31828
    },
    {
      "epoch": 12.716340391530164,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 31829
    },
    {
      "epoch": 12.716739912105474,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 31830
    },
    {
      "epoch": 12.717139432680783,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 31831
    },
    {
      "epoch": 12.717538953256092,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 31832
    },
    {
      "epoch": 12.717938473831403,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 31833
    },
    {
      "epoch": 12.718337994406712,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 31834
    },
    {
      "epoch": 12.718737514982022,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 31835
    },
    {
      "epoch": 12.719137035557331,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 31836
    },
    {
      "epoch": 12.71953655613264,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 31837
    },
    {
      "epoch": 12.71993607670795,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 31838
    },
    {
      "epoch": 12.72033559728326,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 31839
    },
    {
      "epoch": 12.72073511785857,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 31840
    },
    {
      "epoch": 12.72113463843388,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 31841
    },
    {
      "epoch": 12.721534159009188,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 31842
    },
    {
      "epoch": 12.721933679584499,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 31843
    },
    {
      "epoch": 12.722333200159808,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 31844
    },
    {
      "epoch": 12.722732720735118,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 31845
    },
    {
      "epoch": 12.723132241310427,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 31846
    },
    {
      "epoch": 12.723531761885738,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 31847
    },
    {
      "epoch": 12.723931282461047,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 31848
    },
    {
      "epoch": 12.724330803036356,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 31849
    },
    {
      "epoch": 12.724730323611666,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 31850
    },
    {
      "epoch": 12.725129844186975,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 31851
    },
    {
      "epoch": 12.725529364762286,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 31852
    },
    {
      "epoch": 12.725928885337595,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 31853
    },
    {
      "epoch": 12.726328405912904,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 31854
    },
    {
      "epoch": 12.726727926488214,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 31855
    },
    {
      "epoch": 12.727127447063523,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 31856
    },
    {
      "epoch": 12.727526967638834,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 31857
    },
    {
      "epoch": 12.727926488214143,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 31858
    },
    {
      "epoch": 12.728326008789452,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 31859
    },
    {
      "epoch": 12.728725529364763,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 31860
    },
    {
      "epoch": 12.729125049940071,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 31861
    },
    {
      "epoch": 12.729524570515382,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 31862
    },
    {
      "epoch": 12.729924091090691,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 31863
    },
    {
      "epoch": 12.730323611666002,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 31864
    },
    {
      "epoch": 12.73072313224131,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 31865
    },
    {
      "epoch": 12.73112265281662,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 31866
    },
    {
      "epoch": 12.73152217339193,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 31867
    },
    {
      "epoch": 12.731921693967239,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 31868
    },
    {
      "epoch": 12.73232121454255,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 31869
    },
    {
      "epoch": 12.732720735117859,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 31870
    },
    {
      "epoch": 12.733120255693168,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 31871
    },
    {
      "epoch": 12.733519776268478,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 31872
    },
    {
      "epoch": 12.733919296843787,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 31873
    },
    {
      "epoch": 12.734318817419098,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 31874
    },
    {
      "epoch": 12.734718337994407,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 31875
    },
    {
      "epoch": 12.735117858569716,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 31876
    },
    {
      "epoch": 12.735517379145026,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 31877
    },
    {
      "epoch": 12.735916899720335,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 31878
    },
    {
      "epoch": 12.736316420295646,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 31879
    },
    {
      "epoch": 12.736715940870955,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 31880
    },
    {
      "epoch": 12.737115461446265,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 31881
    },
    {
      "epoch": 12.737514982021574,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 31882
    },
    {
      "epoch": 12.737914502596883,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 31883
    },
    {
      "epoch": 12.738314023172194,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 31884
    },
    {
      "epoch": 12.738713543747503,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 31885
    },
    {
      "epoch": 12.739113064322813,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 31886
    },
    {
      "epoch": 12.739512584898122,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 31887
    },
    {
      "epoch": 12.739912105473431,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 31888
    },
    {
      "epoch": 12.740311626048742,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 31889
    },
    {
      "epoch": 12.74071114662405,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 31890
    },
    {
      "epoch": 12.741110667199361,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 31891
    },
    {
      "epoch": 12.74151018777467,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 31892
    },
    {
      "epoch": 12.74190970834998,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 31893
    },
    {
      "epoch": 12.74230922892529,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 31894
    },
    {
      "epoch": 12.742708749500599,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 31895
    },
    {
      "epoch": 12.74310827007591,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 31896
    },
    {
      "epoch": 12.743507790651218,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 31897
    },
    {
      "epoch": 12.743907311226529,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 31898
    },
    {
      "epoch": 12.744306831801838,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 31899
    },
    {
      "epoch": 12.744706352377147,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 31900
    },
    {
      "epoch": 12.745105872952458,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 31901
    },
    {
      "epoch": 12.745505393527766,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 31902
    },
    {
      "epoch": 12.745904914103077,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 31903
    },
    {
      "epoch": 12.746304434678386,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 31904
    },
    {
      "epoch": 12.746703955253695,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 31905
    },
    {
      "epoch": 12.747103475829006,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 31906
    },
    {
      "epoch": 12.747502996404315,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 31907
    },
    {
      "epoch": 12.747902516979625,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 31908
    },
    {
      "epoch": 12.748302037554934,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 31909
    },
    {
      "epoch": 12.748701558130243,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 31910
    },
    {
      "epoch": 12.749101078705554,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 31911
    },
    {
      "epoch": 12.749500599280863,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 31912
    },
    {
      "epoch": 12.749900119856173,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 31913
    },
    {
      "epoch": 12.750299640431482,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 31914
    },
    {
      "epoch": 12.750699161006793,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 31915
    },
    {
      "epoch": 12.751098681582102,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 31916
    },
    {
      "epoch": 12.75149820215741,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 31917
    },
    {
      "epoch": 12.751897722732721,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 31918
    },
    {
      "epoch": 12.75229724330803,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 31919
    },
    {
      "epoch": 12.752696763883339,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 31920
    },
    {
      "epoch": 12.75309628445865,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 31921
    },
    {
      "epoch": 12.753495805033959,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0074,
      "step": 31922
    },
    {
      "epoch": 12.75389532560927,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 31923
    },
    {
      "epoch": 12.754294846184578,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 31924
    },
    {
      "epoch": 12.754694366759889,
      "grad_norm": 0.0361328125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 31925
    },
    {
      "epoch": 12.755093887335198,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 31926
    },
    {
      "epoch": 12.755493407910507,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 31927
    },
    {
      "epoch": 12.755892928485817,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 31928
    },
    {
      "epoch": 12.756292449061126,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 31929
    },
    {
      "epoch": 12.756691969636437,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 31930
    },
    {
      "epoch": 12.757091490211746,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 31931
    },
    {
      "epoch": 12.757491010787055,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 31932
    },
    {
      "epoch": 12.757890531362365,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 31933
    },
    {
      "epoch": 12.758290051937674,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 31934
    },
    {
      "epoch": 12.758689572512985,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 31935
    },
    {
      "epoch": 12.759089093088294,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 31936
    },
    {
      "epoch": 12.759488613663603,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 31937
    },
    {
      "epoch": 12.759888134238913,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 31938
    },
    {
      "epoch": 12.760287654814222,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 31939
    },
    {
      "epoch": 12.760687175389533,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 31940
    },
    {
      "epoch": 12.761086695964842,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 31941
    },
    {
      "epoch": 12.761486216540153,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 31942
    },
    {
      "epoch": 12.761885737115461,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 31943
    },
    {
      "epoch": 12.76228525769077,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 31944
    },
    {
      "epoch": 12.762684778266081,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 31945
    },
    {
      "epoch": 12.76308429884139,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 31946
    },
    {
      "epoch": 12.7634838194167,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 31947
    },
    {
      "epoch": 12.76388333999201,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 31948
    },
    {
      "epoch": 12.764282860567318,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 31949
    },
    {
      "epoch": 12.764682381142629,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 31950
    },
    {
      "epoch": 12.765081901717938,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 31951
    },
    {
      "epoch": 12.765481422293249,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 31952
    },
    {
      "epoch": 12.765880942868558,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 31953
    },
    {
      "epoch": 12.766280463443866,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 31954
    },
    {
      "epoch": 12.766679984019177,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 31955
    },
    {
      "epoch": 12.767079504594486,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 31956
    },
    {
      "epoch": 12.767479025169797,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 31957
    },
    {
      "epoch": 12.767878545745106,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 31958
    },
    {
      "epoch": 12.768278066320416,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 31959
    },
    {
      "epoch": 12.768677586895725,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 31960
    },
    {
      "epoch": 12.769077107471034,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 31961
    },
    {
      "epoch": 12.769476628046345,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 31962
    },
    {
      "epoch": 12.769876148621654,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 31963
    },
    {
      "epoch": 12.770275669196964,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 31964
    },
    {
      "epoch": 12.770675189772273,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 31965
    },
    {
      "epoch": 12.771074710347582,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 31966
    },
    {
      "epoch": 12.771474230922893,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 31967
    },
    {
      "epoch": 12.771873751498202,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 31968
    },
    {
      "epoch": 12.772273272073512,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 31969
    },
    {
      "epoch": 12.772672792648821,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 31970
    },
    {
      "epoch": 12.77307231322413,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 31971
    },
    {
      "epoch": 12.77347183379944,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 31972
    },
    {
      "epoch": 12.77387135437475,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 31973
    },
    {
      "epoch": 12.77427087495006,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 31974
    },
    {
      "epoch": 12.77467039552537,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 31975
    },
    {
      "epoch": 12.77506991610068,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 31976
    },
    {
      "epoch": 12.775469436675989,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 31977
    },
    {
      "epoch": 12.775868957251298,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 31978
    },
    {
      "epoch": 12.776268477826608,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 31979
    },
    {
      "epoch": 12.776667998401917,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 31980
    },
    {
      "epoch": 12.777067518977228,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 31981
    },
    {
      "epoch": 12.777467039552537,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 31982
    },
    {
      "epoch": 12.777866560127846,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 31983
    },
    {
      "epoch": 12.778266080703157,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 31984
    },
    {
      "epoch": 12.778665601278465,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 31985
    },
    {
      "epoch": 12.779065121853776,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 31986
    },
    {
      "epoch": 12.779464642429085,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 31987
    },
    {
      "epoch": 12.779864163004394,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 31988
    },
    {
      "epoch": 12.780263683579705,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 31989
    },
    {
      "epoch": 12.780663204155013,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 31990
    },
    {
      "epoch": 12.781062724730324,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 31991
    },
    {
      "epoch": 12.781462245305633,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 31992
    },
    {
      "epoch": 12.781861765880944,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 31993
    },
    {
      "epoch": 12.782261286456253,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 31994
    },
    {
      "epoch": 12.782660807031561,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 31995
    },
    {
      "epoch": 12.783060327606872,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 31996
    },
    {
      "epoch": 12.783459848182181,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 31997
    },
    {
      "epoch": 12.783859368757492,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 31998
    },
    {
      "epoch": 12.7842588893328,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 31999
    },
    {
      "epoch": 12.78465840990811,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 32000
    },
    {
      "epoch": 12.78505793048342,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 32001
    },
    {
      "epoch": 12.785457451058729,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 32002
    },
    {
      "epoch": 12.78585697163404,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 32003
    },
    {
      "epoch": 12.786256492209349,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 32004
    },
    {
      "epoch": 12.786656012784658,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 32005
    },
    {
      "epoch": 12.787055533359968,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 32006
    },
    {
      "epoch": 12.787455053935277,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 32007
    },
    {
      "epoch": 12.787854574510588,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 32008
    },
    {
      "epoch": 12.788254095085897,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 32009
    },
    {
      "epoch": 12.788653615661207,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 32010
    },
    {
      "epoch": 12.789053136236516,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 32011
    },
    {
      "epoch": 12.789452656811825,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 32012
    },
    {
      "epoch": 12.789852177387136,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 32013
    },
    {
      "epoch": 12.790251697962445,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 32014
    },
    {
      "epoch": 12.790651218537755,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 32015
    },
    {
      "epoch": 12.791050739113064,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 32016
    },
    {
      "epoch": 12.791450259688373,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 32017
    },
    {
      "epoch": 12.791849780263684,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 32018
    },
    {
      "epoch": 12.792249300838993,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 32019
    },
    {
      "epoch": 12.792648821414303,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 32020
    },
    {
      "epoch": 12.793048341989612,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 32021
    },
    {
      "epoch": 12.793447862564921,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 32022
    },
    {
      "epoch": 12.793847383140232,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 32023
    },
    {
      "epoch": 12.79424690371554,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 32024
    },
    {
      "epoch": 12.794646424290852,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 32025
    },
    {
      "epoch": 12.79504594486616,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 32026
    },
    {
      "epoch": 12.795445465441471,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 32027
    },
    {
      "epoch": 12.79584498601678,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 32028
    },
    {
      "epoch": 12.796244506592089,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 32029
    },
    {
      "epoch": 12.7966440271674,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 32030
    },
    {
      "epoch": 12.797043547742708,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 32031
    },
    {
      "epoch": 12.79744306831802,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 32032
    },
    {
      "epoch": 12.797842588893328,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 32033
    },
    {
      "epoch": 12.798242109468637,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 32034
    },
    {
      "epoch": 12.798641630043948,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 32035
    },
    {
      "epoch": 12.799041150619257,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 32036
    },
    {
      "epoch": 12.799440671194567,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 32037
    },
    {
      "epoch": 12.799840191769876,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 32038
    },
    {
      "epoch": 12.800239712345185,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 32039
    },
    {
      "epoch": 12.800639232920496,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 32040
    },
    {
      "epoch": 12.801038753495805,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 32041
    },
    {
      "epoch": 12.801438274071115,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 32042
    },
    {
      "epoch": 12.801837794646424,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 32043
    },
    {
      "epoch": 12.802237315221735,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 32044
    },
    {
      "epoch": 12.802636835797044,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 32045
    },
    {
      "epoch": 12.803036356372353,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 32046
    },
    {
      "epoch": 12.803435876947663,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 32047
    },
    {
      "epoch": 12.803835397522972,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 32048
    },
    {
      "epoch": 12.804234918098283,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 32049
    },
    {
      "epoch": 12.804634438673592,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 32050
    },
    {
      "epoch": 12.8050339592489,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 32051
    },
    {
      "epoch": 12.805433479824211,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 32052
    },
    {
      "epoch": 12.80583300039952,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 32053
    },
    {
      "epoch": 12.806232520974831,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 32054
    },
    {
      "epoch": 12.80663204155014,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 32055
    },
    {
      "epoch": 12.807031562125449,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 32056
    },
    {
      "epoch": 12.80743108270076,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 32057
    },
    {
      "epoch": 12.807830603276068,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 32058
    },
    {
      "epoch": 12.808230123851379,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 32059
    },
    {
      "epoch": 12.808629644426688,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 32060
    },
    {
      "epoch": 12.809029165001999,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 32061
    },
    {
      "epoch": 12.809428685577307,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 32062
    },
    {
      "epoch": 12.809828206152616,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 32063
    },
    {
      "epoch": 12.810227726727927,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 32064
    },
    {
      "epoch": 12.810627247303236,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 32065
    },
    {
      "epoch": 12.811026767878547,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 32066
    },
    {
      "epoch": 12.811426288453855,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 32067
    },
    {
      "epoch": 12.811825809029164,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 32068
    },
    {
      "epoch": 12.812225329604475,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 32069
    },
    {
      "epoch": 12.812624850179784,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 32070
    },
    {
      "epoch": 12.813024370755095,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 32071
    },
    {
      "epoch": 12.813423891330403,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 32072
    },
    {
      "epoch": 12.813823411905712,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 32073
    },
    {
      "epoch": 12.814222932481023,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 32074
    },
    {
      "epoch": 12.814622453056332,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 32075
    },
    {
      "epoch": 12.815021973631643,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 32076
    },
    {
      "epoch": 12.815421494206952,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 32077
    },
    {
      "epoch": 12.815821014782262,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 32078
    },
    {
      "epoch": 12.816220535357571,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 32079
    },
    {
      "epoch": 12.81662005593288,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 32080
    },
    {
      "epoch": 12.81701957650819,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 32081
    },
    {
      "epoch": 12.8174190970835,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 32082
    },
    {
      "epoch": 12.81781861765881,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 32083
    },
    {
      "epoch": 12.81821813823412,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 32084
    },
    {
      "epoch": 12.818617658809428,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 32085
    },
    {
      "epoch": 12.819017179384739,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 32086
    },
    {
      "epoch": 12.819416699960048,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 32087
    },
    {
      "epoch": 12.819816220535358,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 32088
    },
    {
      "epoch": 12.820215741110667,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 32089
    },
    {
      "epoch": 12.820615261685976,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 32090
    },
    {
      "epoch": 12.821014782261287,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0109,
      "step": 32091
    },
    {
      "epoch": 12.821414302836596,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 32092
    },
    {
      "epoch": 12.821813823411906,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 32093
    },
    {
      "epoch": 12.822213343987215,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 32094
    },
    {
      "epoch": 12.822612864562524,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 32095
    },
    {
      "epoch": 12.823012385137835,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0523,
      "step": 32096
    },
    {
      "epoch": 12.823411905713144,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 32097
    },
    {
      "epoch": 12.823811426288454,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 32098
    },
    {
      "epoch": 12.824210946863763,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 32099
    },
    {
      "epoch": 12.824610467439072,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 32100
    },
    {
      "epoch": 12.825009988014383,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 32101
    },
    {
      "epoch": 12.825409508589692,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 32102
    },
    {
      "epoch": 12.825809029165002,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 32103
    },
    {
      "epoch": 12.826208549740311,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 32104
    },
    {
      "epoch": 12.826608070315622,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 32105
    },
    {
      "epoch": 12.827007590890931,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 32106
    },
    {
      "epoch": 12.82740711146624,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 32107
    },
    {
      "epoch": 12.82780663204155,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 32108
    },
    {
      "epoch": 12.82820615261686,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 32109
    },
    {
      "epoch": 12.82860567319217,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 32110
    },
    {
      "epoch": 12.829005193767479,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 32111
    },
    {
      "epoch": 12.829404714342788,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 32112
    },
    {
      "epoch": 12.829804234918099,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 32113
    },
    {
      "epoch": 12.830203755493407,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 32114
    },
    {
      "epoch": 12.830603276068718,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 32115
    },
    {
      "epoch": 12.831002796644027,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 32116
    },
    {
      "epoch": 12.831402317219336,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 32117
    },
    {
      "epoch": 12.831801837794647,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 32118
    },
    {
      "epoch": 12.832201358369955,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 32119
    },
    {
      "epoch": 12.832600878945266,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 32120
    },
    {
      "epoch": 12.833000399520575,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 32121
    },
    {
      "epoch": 12.833399920095886,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 32122
    },
    {
      "epoch": 12.833799440671195,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 32123
    },
    {
      "epoch": 12.834198961246504,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 32124
    },
    {
      "epoch": 12.834598481821814,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 32125
    },
    {
      "epoch": 12.834998002397123,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 32126
    },
    {
      "epoch": 12.835397522972434,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 32127
    },
    {
      "epoch": 12.835797043547743,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 32128
    },
    {
      "epoch": 12.836196564123052,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 32129
    },
    {
      "epoch": 12.836596084698362,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 32130
    },
    {
      "epoch": 12.836995605273671,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 32131
    },
    {
      "epoch": 12.837395125848982,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 32132
    },
    {
      "epoch": 12.83779464642429,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 32133
    },
    {
      "epoch": 12.8381941669996,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 32134
    },
    {
      "epoch": 12.83859368757491,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 32135
    },
    {
      "epoch": 12.83899320815022,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 32136
    },
    {
      "epoch": 12.83939272872553,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 32137
    },
    {
      "epoch": 12.839792249300839,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 32138
    },
    {
      "epoch": 12.84019176987615,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 32139
    },
    {
      "epoch": 12.840591290451458,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 32140
    },
    {
      "epoch": 12.840990811026767,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 32141
    },
    {
      "epoch": 12.841390331602078,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 32142
    },
    {
      "epoch": 12.841789852177387,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 32143
    },
    {
      "epoch": 12.842189372752697,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 32144
    },
    {
      "epoch": 12.842588893328006,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 32145
    },
    {
      "epoch": 12.842988413903315,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 32146
    },
    {
      "epoch": 12.843387934478626,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 32147
    },
    {
      "epoch": 12.843787455053935,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 32148
    },
    {
      "epoch": 12.844186975629245,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 32149
    },
    {
      "epoch": 12.844586496204554,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 32150
    },
    {
      "epoch": 12.844986016779863,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 32151
    },
    {
      "epoch": 12.845385537355174,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 32152
    },
    {
      "epoch": 12.845785057930483,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 32153
    },
    {
      "epoch": 12.846184578505794,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 32154
    },
    {
      "epoch": 12.846584099081102,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 32155
    },
    {
      "epoch": 12.846983619656413,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 32156
    },
    {
      "epoch": 12.847383140231722,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0519,
      "step": 32157
    },
    {
      "epoch": 12.847782660807031,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 32158
    },
    {
      "epoch": 12.848182181382342,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 32159
    },
    {
      "epoch": 12.84858170195765,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 32160
    },
    {
      "epoch": 12.848981222532961,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 32161
    },
    {
      "epoch": 12.84938074310827,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 32162
    },
    {
      "epoch": 12.849780263683579,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 32163
    },
    {
      "epoch": 12.85017978425889,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 32164
    },
    {
      "epoch": 12.850579304834199,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 32165
    },
    {
      "epoch": 12.85097882540951,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 32166
    },
    {
      "epoch": 12.851378345984818,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 32167
    },
    {
      "epoch": 12.851777866560127,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 32168
    },
    {
      "epoch": 12.852177387135438,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 32169
    },
    {
      "epoch": 12.852576907710747,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 32170
    },
    {
      "epoch": 12.852976428286057,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 32171
    },
    {
      "epoch": 12.853375948861366,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 32172
    },
    {
      "epoch": 12.853775469436677,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 32173
    },
    {
      "epoch": 12.854174990011986,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 32174
    },
    {
      "epoch": 12.854574510587295,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 32175
    },
    {
      "epoch": 12.854974031162605,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 32176
    },
    {
      "epoch": 12.855373551737914,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 32177
    },
    {
      "epoch": 12.855773072313225,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 32178
    },
    {
      "epoch": 12.856172592888534,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 32179
    },
    {
      "epoch": 12.856572113463843,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 32180
    },
    {
      "epoch": 12.856971634039153,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 32181
    },
    {
      "epoch": 12.857371154614462,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 32182
    },
    {
      "epoch": 12.857770675189773,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 32183
    },
    {
      "epoch": 12.858170195765082,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 32184
    },
    {
      "epoch": 12.85856971634039,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 32185
    },
    {
      "epoch": 12.858969236915701,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 32186
    },
    {
      "epoch": 12.85936875749101,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 32187
    },
    {
      "epoch": 12.859768278066321,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 32188
    },
    {
      "epoch": 12.86016779864163,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 32189
    },
    {
      "epoch": 12.86056731921694,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 32190
    },
    {
      "epoch": 12.86096683979225,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 32191
    },
    {
      "epoch": 12.861366360367558,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 32192
    },
    {
      "epoch": 12.861765880942869,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 32193
    },
    {
      "epoch": 12.862165401518178,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 32194
    },
    {
      "epoch": 12.862564922093489,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 32195
    },
    {
      "epoch": 12.862964442668797,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 32196
    },
    {
      "epoch": 12.863363963244106,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 32197
    },
    {
      "epoch": 12.863763483819417,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 32198
    },
    {
      "epoch": 12.864163004394726,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 32199
    },
    {
      "epoch": 12.864562524970037,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 32200
    },
    {
      "epoch": 12.864962045545346,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 32201
    },
    {
      "epoch": 12.865361566120654,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 32202
    },
    {
      "epoch": 12.865761086695965,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 32203
    },
    {
      "epoch": 12.866160607271274,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 32204
    },
    {
      "epoch": 12.866560127846585,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 32205
    },
    {
      "epoch": 12.866959648421894,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 32206
    },
    {
      "epoch": 12.867359168997204,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 32207
    },
    {
      "epoch": 12.867758689572513,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 32208
    },
    {
      "epoch": 12.868158210147822,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 32209
    },
    {
      "epoch": 12.868557730723133,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 32210
    },
    {
      "epoch": 12.868957251298442,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 32211
    },
    {
      "epoch": 12.869356771873752,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 32212
    },
    {
      "epoch": 12.869756292449061,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 32213
    },
    {
      "epoch": 12.87015581302437,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 32214
    },
    {
      "epoch": 12.87055533359968,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 32215
    },
    {
      "epoch": 12.87095485417499,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 32216
    },
    {
      "epoch": 12.8713543747503,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 32217
    },
    {
      "epoch": 12.87175389532561,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 32218
    },
    {
      "epoch": 12.872153415900918,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 32219
    },
    {
      "epoch": 12.872552936476229,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 32220
    },
    {
      "epoch": 12.872952457051538,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 32221
    },
    {
      "epoch": 12.873351977626848,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 32222
    },
    {
      "epoch": 12.873751498202157,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 32223
    },
    {
      "epoch": 12.874151018777468,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 32224
    },
    {
      "epoch": 12.874550539352777,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 32225
    },
    {
      "epoch": 12.874950059928086,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 32226
    },
    {
      "epoch": 12.875349580503396,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 32227
    },
    {
      "epoch": 12.875749101078705,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 32228
    },
    {
      "epoch": 12.876148621654016,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 32229
    },
    {
      "epoch": 12.876548142229325,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 32230
    },
    {
      "epoch": 12.876947662804634,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 32231
    },
    {
      "epoch": 12.877347183379944,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 32232
    },
    {
      "epoch": 12.877746703955253,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 32233
    },
    {
      "epoch": 12.878146224530564,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 32234
    },
    {
      "epoch": 12.878545745105873,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 32235
    },
    {
      "epoch": 12.878945265681182,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 32236
    },
    {
      "epoch": 12.879344786256492,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 32237
    },
    {
      "epoch": 12.879744306831801,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 32238
    },
    {
      "epoch": 12.880143827407112,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 32239
    },
    {
      "epoch": 12.880543347982421,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 32240
    },
    {
      "epoch": 12.880942868557732,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 32241
    },
    {
      "epoch": 12.88134238913304,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 32242
    },
    {
      "epoch": 12.88174190970835,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 32243
    },
    {
      "epoch": 12.88214143028366,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 32244
    },
    {
      "epoch": 12.882540950858969,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 32245
    },
    {
      "epoch": 12.88294047143428,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 32246
    },
    {
      "epoch": 12.883339992009589,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 32247
    },
    {
      "epoch": 12.883739512584897,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 32248
    },
    {
      "epoch": 12.884139033160208,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 32249
    },
    {
      "epoch": 12.884538553735517,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 32250
    },
    {
      "epoch": 12.884938074310828,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 32251
    },
    {
      "epoch": 12.885337594886137,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 32252
    },
    {
      "epoch": 12.885737115461446,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 32253
    },
    {
      "epoch": 12.886136636036756,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 32254
    },
    {
      "epoch": 12.886536156612065,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 32255
    },
    {
      "epoch": 12.886935677187376,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 32256
    },
    {
      "epoch": 12.887335197762685,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 32257
    },
    {
      "epoch": 12.887734718337995,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 32258
    },
    {
      "epoch": 12.888134238913304,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 32259
    },
    {
      "epoch": 12.888533759488613,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 32260
    },
    {
      "epoch": 12.888933280063924,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 32261
    },
    {
      "epoch": 12.889332800639233,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 32262
    },
    {
      "epoch": 12.889732321214543,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 32263
    },
    {
      "epoch": 12.890131841789852,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0115,
      "step": 32264
    },
    {
      "epoch": 12.890531362365161,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 32265
    },
    {
      "epoch": 12.890930882940472,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 32266
    },
    {
      "epoch": 12.89133040351578,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 32267
    },
    {
      "epoch": 12.891729924091091,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 32268
    },
    {
      "epoch": 12.8921294446664,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 32269
    },
    {
      "epoch": 12.89252896524171,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 32270
    },
    {
      "epoch": 12.89292848581702,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 32271
    },
    {
      "epoch": 12.893328006392329,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 32272
    },
    {
      "epoch": 12.89372752696764,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 32273
    },
    {
      "epoch": 12.894127047542948,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 32274
    },
    {
      "epoch": 12.894526568118257,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 32275
    },
    {
      "epoch": 12.894926088693568,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 32276
    },
    {
      "epoch": 12.895325609268877,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 32277
    },
    {
      "epoch": 12.895725129844188,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 32278
    },
    {
      "epoch": 12.896124650419496,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 32279
    },
    {
      "epoch": 12.896524170994805,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 32280
    },
    {
      "epoch": 12.896923691570116,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 32281
    },
    {
      "epoch": 12.897323212145425,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 32282
    },
    {
      "epoch": 12.897722732720736,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 32283
    },
    {
      "epoch": 12.898122253296044,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 32284
    },
    {
      "epoch": 12.898521773871355,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 32285
    },
    {
      "epoch": 12.898921294446664,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 32286
    },
    {
      "epoch": 12.899320815021973,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 32287
    },
    {
      "epoch": 12.899720335597284,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 32288
    },
    {
      "epoch": 12.900119856172592,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 32289
    },
    {
      "epoch": 12.900519376747903,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 32290
    },
    {
      "epoch": 12.900918897323212,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 32291
    },
    {
      "epoch": 12.901318417898521,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 32292
    },
    {
      "epoch": 12.901717938473832,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 32293
    },
    {
      "epoch": 12.90211745904914,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 32294
    },
    {
      "epoch": 12.902516979624451,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 32295
    },
    {
      "epoch": 12.90291650019976,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 32296
    },
    {
      "epoch": 12.903316020775069,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 32297
    },
    {
      "epoch": 12.90371554135038,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 32298
    },
    {
      "epoch": 12.904115061925689,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 32299
    },
    {
      "epoch": 12.904514582501,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 32300
    },
    {
      "epoch": 12.904914103076308,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 32301
    },
    {
      "epoch": 12.905313623651619,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 32302
    },
    {
      "epoch": 12.905713144226928,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 32303
    },
    {
      "epoch": 12.906112664802237,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 32304
    },
    {
      "epoch": 12.906512185377547,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 32305
    },
    {
      "epoch": 12.906911705952856,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 32306
    },
    {
      "epoch": 12.907311226528167,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 32307
    },
    {
      "epoch": 12.907710747103476,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 32308
    },
    {
      "epoch": 12.908110267678785,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 32309
    },
    {
      "epoch": 12.908509788254095,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 32310
    },
    {
      "epoch": 12.908909308829404,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 32311
    },
    {
      "epoch": 12.909308829404715,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 32312
    },
    {
      "epoch": 12.909708349980024,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 32313
    },
    {
      "epoch": 12.910107870555333,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 32314
    },
    {
      "epoch": 12.910507391130643,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 32315
    },
    {
      "epoch": 12.910906911705952,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 32316
    },
    {
      "epoch": 12.911306432281263,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 32317
    },
    {
      "epoch": 12.911705952856572,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 32318
    },
    {
      "epoch": 12.912105473431883,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 32319
    },
    {
      "epoch": 12.912504994007191,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 32320
    },
    {
      "epoch": 12.9129045145825,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 32321
    },
    {
      "epoch": 12.913304035157811,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 32322
    },
    {
      "epoch": 12.91370355573312,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 32323
    },
    {
      "epoch": 12.91410307630843,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 32324
    },
    {
      "epoch": 12.91450259688374,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 32325
    },
    {
      "epoch": 12.914902117459048,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 32326
    },
    {
      "epoch": 12.915301638034359,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 32327
    },
    {
      "epoch": 12.915701158609668,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 32328
    },
    {
      "epoch": 12.916100679184979,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 32329
    },
    {
      "epoch": 12.916500199760288,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 32330
    },
    {
      "epoch": 12.916899720335596,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 32331
    },
    {
      "epoch": 12.917299240910907,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 32332
    },
    {
      "epoch": 12.917698761486216,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 32333
    },
    {
      "epoch": 12.918098282061527,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 32334
    },
    {
      "epoch": 12.918497802636836,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 32335
    },
    {
      "epoch": 12.918897323212146,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 32336
    },
    {
      "epoch": 12.919296843787455,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 32337
    },
    {
      "epoch": 12.919696364362764,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 32338
    },
    {
      "epoch": 12.920095884938075,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 32339
    },
    {
      "epoch": 12.920495405513384,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 32340
    },
    {
      "epoch": 12.920894926088694,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 32341
    },
    {
      "epoch": 12.921294446664003,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 32342
    },
    {
      "epoch": 12.921693967239312,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 32343
    },
    {
      "epoch": 12.922093487814623,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 32344
    },
    {
      "epoch": 12.922493008389932,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 32345
    },
    {
      "epoch": 12.922892528965242,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 32346
    },
    {
      "epoch": 12.923292049540551,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 32347
    },
    {
      "epoch": 12.92369157011586,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 32348
    },
    {
      "epoch": 12.92409109069117,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 32349
    },
    {
      "epoch": 12.92449061126648,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 32350
    },
    {
      "epoch": 12.92489013184179,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 32351
    },
    {
      "epoch": 12.9252896524171,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 32352
    },
    {
      "epoch": 12.92568917299241,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 32353
    },
    {
      "epoch": 12.926088693567719,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 32354
    },
    {
      "epoch": 12.926488214143028,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 32355
    },
    {
      "epoch": 12.926887734718338,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 32356
    },
    {
      "epoch": 12.927287255293647,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 32357
    },
    {
      "epoch": 12.927686775868958,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 32358
    },
    {
      "epoch": 12.928086296444267,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 32359
    },
    {
      "epoch": 12.928485817019576,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 32360
    },
    {
      "epoch": 12.928885337594886,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 32361
    },
    {
      "epoch": 12.929284858170195,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 32362
    },
    {
      "epoch": 12.929684378745506,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 32363
    },
    {
      "epoch": 12.930083899320815,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 32364
    },
    {
      "epoch": 12.930483419896124,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 32365
    },
    {
      "epoch": 12.930882940471434,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 32366
    },
    {
      "epoch": 12.931282461046743,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 32367
    },
    {
      "epoch": 12.931681981622054,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 32368
    },
    {
      "epoch": 12.932081502197363,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 32369
    },
    {
      "epoch": 12.932481022772674,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 32370
    },
    {
      "epoch": 12.932880543347983,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 32371
    },
    {
      "epoch": 12.933280063923291,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 32372
    },
    {
      "epoch": 12.933679584498602,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 32373
    },
    {
      "epoch": 12.934079105073911,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 32374
    },
    {
      "epoch": 12.934478625649222,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 32375
    },
    {
      "epoch": 12.93487814622453,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 32376
    },
    {
      "epoch": 12.93527766679984,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 32377
    },
    {
      "epoch": 12.93567718737515,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 32378
    },
    {
      "epoch": 12.936076707950459,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 32379
    },
    {
      "epoch": 12.93647622852577,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 32380
    },
    {
      "epoch": 12.936875749101079,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 32381
    },
    {
      "epoch": 12.937275269676388,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 32382
    },
    {
      "epoch": 12.937674790251698,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 32383
    },
    {
      "epoch": 12.938074310827007,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 32384
    },
    {
      "epoch": 12.938473831402318,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 32385
    },
    {
      "epoch": 12.938873351977627,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 32386
    },
    {
      "epoch": 12.939272872552937,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 32387
    },
    {
      "epoch": 12.939672393128246,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 32388
    },
    {
      "epoch": 12.940071913703555,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 32389
    },
    {
      "epoch": 12.940471434278866,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 32390
    },
    {
      "epoch": 12.940870954854175,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 32391
    },
    {
      "epoch": 12.941270475429485,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 32392
    },
    {
      "epoch": 12.941669996004794,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 32393
    },
    {
      "epoch": 12.942069516580103,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 32394
    },
    {
      "epoch": 12.942469037155414,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 32395
    },
    {
      "epoch": 12.942868557730723,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 32396
    },
    {
      "epoch": 12.943268078306033,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 32397
    },
    {
      "epoch": 12.943667598881342,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 32398
    },
    {
      "epoch": 12.944067119456651,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 32399
    },
    {
      "epoch": 12.944466640031962,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 32400
    },
    {
      "epoch": 12.94486616060727,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 32401
    },
    {
      "epoch": 12.945265681182581,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 32402
    },
    {
      "epoch": 12.94566520175789,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 32403
    },
    {
      "epoch": 12.946064722333201,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 32404
    },
    {
      "epoch": 12.94646424290851,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 32405
    },
    {
      "epoch": 12.946863763483819,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 32406
    },
    {
      "epoch": 12.94726328405913,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 32407
    },
    {
      "epoch": 12.947662804634438,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 32408
    },
    {
      "epoch": 12.948062325209749,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 32409
    },
    {
      "epoch": 12.948461845785058,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 32410
    },
    {
      "epoch": 12.948861366360367,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 32411
    },
    {
      "epoch": 12.949260886935678,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 32412
    },
    {
      "epoch": 12.949660407510986,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 32413
    },
    {
      "epoch": 12.950059928086297,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 32414
    },
    {
      "epoch": 12.950459448661606,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 32415
    },
    {
      "epoch": 12.950858969236915,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 32416
    },
    {
      "epoch": 12.951258489812226,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 32417
    },
    {
      "epoch": 12.951658010387535,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 32418
    },
    {
      "epoch": 12.952057530962845,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 32419
    },
    {
      "epoch": 12.952457051538154,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 32420
    },
    {
      "epoch": 12.952856572113465,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 32421
    },
    {
      "epoch": 12.953256092688774,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 32422
    },
    {
      "epoch": 12.953655613264083,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 32423
    },
    {
      "epoch": 12.954055133839393,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 32424
    },
    {
      "epoch": 12.954454654414702,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 32425
    },
    {
      "epoch": 12.954854174990013,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 32426
    },
    {
      "epoch": 12.955253695565322,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 32427
    },
    {
      "epoch": 12.95565321614063,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 32428
    },
    {
      "epoch": 12.956052736715941,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 32429
    },
    {
      "epoch": 12.95645225729125,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 32430
    },
    {
      "epoch": 12.95685177786656,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 32431
    },
    {
      "epoch": 12.95725129844187,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 32432
    },
    {
      "epoch": 12.957650819017179,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 32433
    },
    {
      "epoch": 12.95805033959249,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 32434
    },
    {
      "epoch": 12.958449860167798,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 32435
    },
    {
      "epoch": 12.958849380743109,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 32436
    },
    {
      "epoch": 12.959248901318418,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 32437
    },
    {
      "epoch": 12.959648421893728,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 32438
    },
    {
      "epoch": 12.960047942469037,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 32439
    },
    {
      "epoch": 12.960447463044346,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 32440
    },
    {
      "epoch": 12.960846983619657,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 32441
    },
    {
      "epoch": 12.961246504194966,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 32442
    },
    {
      "epoch": 12.961646024770276,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 32443
    },
    {
      "epoch": 12.962045545345585,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 32444
    },
    {
      "epoch": 12.962445065920894,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 32445
    },
    {
      "epoch": 12.962844586496205,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 32446
    },
    {
      "epoch": 12.963244107071514,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 32447
    },
    {
      "epoch": 12.963643627646825,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 32448
    },
    {
      "epoch": 12.964043148222133,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 32449
    },
    {
      "epoch": 12.964442668797442,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 32450
    },
    {
      "epoch": 12.964842189372753,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 32451
    },
    {
      "epoch": 12.965241709948062,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0121,
      "step": 32452
    },
    {
      "epoch": 12.965641230523373,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 32453
    },
    {
      "epoch": 12.966040751098681,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 32454
    },
    {
      "epoch": 12.96644027167399,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 32455
    },
    {
      "epoch": 12.966839792249301,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 32456
    },
    {
      "epoch": 12.96723931282461,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 32457
    },
    {
      "epoch": 12.96763883339992,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 32458
    },
    {
      "epoch": 12.96803835397523,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 32459
    },
    {
      "epoch": 12.968437874550538,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0107,
      "step": 32460
    },
    {
      "epoch": 12.968837395125849,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 32461
    },
    {
      "epoch": 12.969236915701158,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 32462
    },
    {
      "epoch": 12.969636436276469,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 32463
    },
    {
      "epoch": 12.970035956851778,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 32464
    },
    {
      "epoch": 12.970435477427088,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 32465
    },
    {
      "epoch": 12.970834998002397,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 32466
    },
    {
      "epoch": 12.971234518577706,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 32467
    },
    {
      "epoch": 12.971634039153017,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 32468
    },
    {
      "epoch": 12.972033559728326,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 32469
    },
    {
      "epoch": 12.972433080303636,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 32470
    },
    {
      "epoch": 12.972832600878945,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 32471
    },
    {
      "epoch": 12.973232121454254,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 32472
    },
    {
      "epoch": 12.973631642029565,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 32473
    },
    {
      "epoch": 12.974031162604874,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 32474
    },
    {
      "epoch": 12.974430683180184,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0116,
      "step": 32475
    },
    {
      "epoch": 12.974830203755493,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 32476
    },
    {
      "epoch": 12.975229724330802,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 32477
    },
    {
      "epoch": 12.975629244906113,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 32478
    },
    {
      "epoch": 12.976028765481422,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 32479
    },
    {
      "epoch": 12.976428286056732,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 32480
    },
    {
      "epoch": 12.976827806632041,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 32481
    },
    {
      "epoch": 12.977227327207352,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 32482
    },
    {
      "epoch": 12.97762684778266,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 32483
    },
    {
      "epoch": 12.97802636835797,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 32484
    },
    {
      "epoch": 12.97842588893328,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 32485
    },
    {
      "epoch": 12.97882540950859,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 32486
    },
    {
      "epoch": 12.9792249300839,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0111,
      "step": 32487
    },
    {
      "epoch": 12.979624450659209,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 32488
    },
    {
      "epoch": 12.980023971234518,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 32489
    },
    {
      "epoch": 12.980423491809828,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 32490
    },
    {
      "epoch": 12.980823012385137,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 32491
    },
    {
      "epoch": 12.981222532960448,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 32492
    },
    {
      "epoch": 12.981622053535757,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 32493
    },
    {
      "epoch": 12.982021574111066,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 32494
    },
    {
      "epoch": 12.982421094686377,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 32495
    },
    {
      "epoch": 12.982820615261685,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 32496
    },
    {
      "epoch": 12.983220135836996,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 32497
    },
    {
      "epoch": 12.983619656412305,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 32498
    },
    {
      "epoch": 12.984019176987616,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 32499
    },
    {
      "epoch": 12.984418697562925,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 32500
    },
    {
      "epoch": 12.984818218138233,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 32501
    },
    {
      "epoch": 12.985217738713544,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 32502
    },
    {
      "epoch": 12.985617259288853,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 32503
    },
    {
      "epoch": 12.986016779864164,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 32504
    },
    {
      "epoch": 12.986416300439473,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 32505
    },
    {
      "epoch": 12.986815821014781,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 32506
    },
    {
      "epoch": 12.987215341590092,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 32507
    },
    {
      "epoch": 12.987614862165401,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 32508
    },
    {
      "epoch": 12.988014382740712,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 32509
    },
    {
      "epoch": 12.98841390331602,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 32510
    },
    {
      "epoch": 12.98881342389133,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 32511
    },
    {
      "epoch": 12.98921294446664,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 32512
    },
    {
      "epoch": 12.989612465041949,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 32513
    },
    {
      "epoch": 12.99001198561726,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 32514
    },
    {
      "epoch": 12.990411506192569,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 32515
    },
    {
      "epoch": 12.99081102676788,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 32516
    },
    {
      "epoch": 12.991210547343188,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 32517
    },
    {
      "epoch": 12.991610067918497,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 32518
    },
    {
      "epoch": 12.992009588493808,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 32519
    },
    {
      "epoch": 12.992409109069117,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 32520
    },
    {
      "epoch": 12.992808629644427,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 32521
    },
    {
      "epoch": 12.993208150219736,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 32522
    },
    {
      "epoch": 12.993607670795045,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 32523
    },
    {
      "epoch": 12.994007191370356,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 32524
    },
    {
      "epoch": 12.994406711945665,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 32525
    },
    {
      "epoch": 12.994806232520975,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 32526
    },
    {
      "epoch": 12.995205753096284,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 32527
    },
    {
      "epoch": 12.995605273671593,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 32528
    },
    {
      "epoch": 12.996004794246904,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 32529
    },
    {
      "epoch": 12.996404314822213,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 32530
    },
    {
      "epoch": 12.996803835397523,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 32531
    },
    {
      "epoch": 12.997203355972832,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 32532
    },
    {
      "epoch": 12.997602876548143,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 32533
    },
    {
      "epoch": 12.998002397123452,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 32534
    },
    {
      "epoch": 12.99840191769876,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 32535
    },
    {
      "epoch": 12.998801438274072,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 32536
    },
    {
      "epoch": 12.99920095884938,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 32537
    },
    {
      "epoch": 12.999600479424691,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 32538
    },
    {
      "epoch": 13.0,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 32539
    },
    {
      "epoch": 13.000399520575309,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 32540
    },
    {
      "epoch": 13.00079904115062,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 32541
    },
    {
      "epoch": 13.001198561725928,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 32542
    },
    {
      "epoch": 13.00159808230124,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 32543
    },
    {
      "epoch": 13.001997602876548,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 32544
    },
    {
      "epoch": 13.002397123451857,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 32545
    },
    {
      "epoch": 13.002796644027168,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 32546
    },
    {
      "epoch": 13.003196164602477,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 32547
    },
    {
      "epoch": 13.003595685177787,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 32548
    },
    {
      "epoch": 13.003995205753096,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 32549
    },
    {
      "epoch": 13.004394726328407,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 32550
    },
    {
      "epoch": 13.004794246903716,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 32551
    },
    {
      "epoch": 13.005193767479025,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 32552
    },
    {
      "epoch": 13.005593288054335,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 32553
    },
    {
      "epoch": 13.005992808629644,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 32554
    },
    {
      "epoch": 13.006392329204955,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 32555
    },
    {
      "epoch": 13.006791849780264,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 32556
    },
    {
      "epoch": 13.007191370355573,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 32557
    },
    {
      "epoch": 13.007590890930883,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0462,
      "step": 32558
    },
    {
      "epoch": 13.007990411506192,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 32559
    },
    {
      "epoch": 13.008389932081503,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 32560
    },
    {
      "epoch": 13.008789452656812,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 32561
    },
    {
      "epoch": 13.00918897323212,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 32562
    },
    {
      "epoch": 13.009588493807431,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 32563
    },
    {
      "epoch": 13.00998801438274,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 32564
    },
    {
      "epoch": 13.010387534958051,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 32565
    },
    {
      "epoch": 13.01078705553336,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 32566
    },
    {
      "epoch": 13.01118657610867,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 32567
    },
    {
      "epoch": 13.01158609668398,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 32568
    },
    {
      "epoch": 13.011985617259288,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 32569
    },
    {
      "epoch": 13.012385137834599,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 32570
    },
    {
      "epoch": 13.012784658409908,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 32571
    },
    {
      "epoch": 13.013184178985219,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 32572
    },
    {
      "epoch": 13.013583699560527,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 32573
    },
    {
      "epoch": 13.013983220135836,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 32574
    },
    {
      "epoch": 13.014382740711147,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 32575
    },
    {
      "epoch": 13.014782261286456,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 32576
    },
    {
      "epoch": 13.015181781861767,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 32577
    },
    {
      "epoch": 13.015581302437075,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 32578
    },
    {
      "epoch": 13.015980823012384,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 32579
    },
    {
      "epoch": 13.016380343587695,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 32580
    },
    {
      "epoch": 13.016779864163004,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 32581
    },
    {
      "epoch": 13.017179384738315,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 32582
    },
    {
      "epoch": 13.017578905313623,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 32583
    },
    {
      "epoch": 13.017978425888934,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 32584
    },
    {
      "epoch": 13.018377946464243,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 32585
    },
    {
      "epoch": 13.018777467039552,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 32586
    },
    {
      "epoch": 13.019176987614863,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 32587
    },
    {
      "epoch": 13.019576508190172,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 32588
    },
    {
      "epoch": 13.019976028765482,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0107,
      "step": 32589
    },
    {
      "epoch": 13.020375549340791,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 32590
    },
    {
      "epoch": 13.0207750699161,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 32591
    },
    {
      "epoch": 13.02117459049141,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 32592
    },
    {
      "epoch": 13.02157411106672,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 32593
    },
    {
      "epoch": 13.02197363164203,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 32594
    },
    {
      "epoch": 13.02237315221734,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 32595
    },
    {
      "epoch": 13.022772672792648,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 32596
    },
    {
      "epoch": 13.023172193367959,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 32597
    },
    {
      "epoch": 13.023571713943268,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 32598
    },
    {
      "epoch": 13.023971234518578,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 32599
    },
    {
      "epoch": 13.024370755093887,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 32600
    },
    {
      "epoch": 13.024770275669196,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 32601
    },
    {
      "epoch": 13.025169796244507,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 32602
    },
    {
      "epoch": 13.025569316819816,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 32603
    },
    {
      "epoch": 13.025968837395126,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 32604
    },
    {
      "epoch": 13.026368357970435,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 32605
    },
    {
      "epoch": 13.026767878545746,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 32606
    },
    {
      "epoch": 13.027167399121055,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 32607
    },
    {
      "epoch": 13.027566919696364,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 32608
    },
    {
      "epoch": 13.027966440271674,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 32609
    },
    {
      "epoch": 13.028365960846983,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 32610
    },
    {
      "epoch": 13.028765481422294,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 32611
    },
    {
      "epoch": 13.029165001997603,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 32612
    },
    {
      "epoch": 13.029564522572912,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 32613
    },
    {
      "epoch": 13.029964043148222,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 32614
    },
    {
      "epoch": 13.030363563723531,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 32615
    },
    {
      "epoch": 13.030763084298842,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 32616
    },
    {
      "epoch": 13.031162604874151,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 32617
    },
    {
      "epoch": 13.03156212544946,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 32618
    },
    {
      "epoch": 13.03196164602477,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 32619
    },
    {
      "epoch": 13.03236116660008,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 32620
    },
    {
      "epoch": 13.03276068717539,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 32621
    },
    {
      "epoch": 13.033160207750699,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 32622
    },
    {
      "epoch": 13.03355972832601,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 32623
    },
    {
      "epoch": 13.033959248901319,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 32624
    },
    {
      "epoch": 13.034358769476627,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 32625
    },
    {
      "epoch": 13.034758290051938,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 32626
    },
    {
      "epoch": 13.035157810627247,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 32627
    },
    {
      "epoch": 13.035557331202558,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 32628
    },
    {
      "epoch": 13.035956851777867,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 32629
    },
    {
      "epoch": 13.036356372353175,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 32630
    },
    {
      "epoch": 13.036755892928486,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 32631
    },
    {
      "epoch": 13.037155413503795,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 32632
    },
    {
      "epoch": 13.037554934079106,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 32633
    },
    {
      "epoch": 13.037954454654415,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 32634
    },
    {
      "epoch": 13.038353975229724,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 32635
    },
    {
      "epoch": 13.038753495805034,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 32636
    },
    {
      "epoch": 13.039153016380343,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 32637
    },
    {
      "epoch": 13.039552536955654,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 32638
    },
    {
      "epoch": 13.039952057530963,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 32639
    },
    {
      "epoch": 13.040351578106273,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 32640
    },
    {
      "epoch": 13.040751098681582,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 32641
    },
    {
      "epoch": 13.041150619256891,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 32642
    },
    {
      "epoch": 13.041550139832202,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 32643
    },
    {
      "epoch": 13.04194966040751,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 32644
    },
    {
      "epoch": 13.042349180982821,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 32645
    },
    {
      "epoch": 13.04274870155813,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 32646
    },
    {
      "epoch": 13.04314822213344,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 32647
    },
    {
      "epoch": 13.04354774270875,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 32648
    },
    {
      "epoch": 13.043947263284059,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 32649
    },
    {
      "epoch": 13.04434678385937,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 32650
    },
    {
      "epoch": 13.044746304434678,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 32651
    },
    {
      "epoch": 13.045145825009987,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 32652
    },
    {
      "epoch": 13.045545345585298,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 32653
    },
    {
      "epoch": 13.045944866160607,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 32654
    },
    {
      "epoch": 13.046344386735917,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 32655
    },
    {
      "epoch": 13.046743907311226,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 32656
    },
    {
      "epoch": 13.047143427886537,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 32657
    },
    {
      "epoch": 13.047542948461846,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 32658
    },
    {
      "epoch": 13.047942469037155,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 32659
    },
    {
      "epoch": 13.048341989612465,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 32660
    },
    {
      "epoch": 13.048741510187774,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 32661
    },
    {
      "epoch": 13.049141030763085,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 32662
    },
    {
      "epoch": 13.049540551338394,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 32663
    },
    {
      "epoch": 13.049940071913703,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 32664
    },
    {
      "epoch": 13.050339592489014,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 32665
    },
    {
      "epoch": 13.050739113064322,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 32666
    },
    {
      "epoch": 13.051138633639633,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 32667
    },
    {
      "epoch": 13.051538154214942,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 32668
    },
    {
      "epoch": 13.051937674790251,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 32669
    },
    {
      "epoch": 13.052337195365562,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 32670
    },
    {
      "epoch": 13.05273671594087,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 32671
    },
    {
      "epoch": 13.053136236516181,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 32672
    },
    {
      "epoch": 13.05353575709149,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 32673
    },
    {
      "epoch": 13.053935277666799,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 32674
    },
    {
      "epoch": 13.05433479824211,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 32675
    },
    {
      "epoch": 13.054734318817419,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 32676
    },
    {
      "epoch": 13.05513383939273,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 32677
    },
    {
      "epoch": 13.055533359968038,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 32678
    },
    {
      "epoch": 13.055932880543349,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 32679
    },
    {
      "epoch": 13.056332401118658,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 32680
    },
    {
      "epoch": 13.056731921693967,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 32681
    },
    {
      "epoch": 13.057131442269277,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 32682
    },
    {
      "epoch": 13.057530962844586,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 32683
    },
    {
      "epoch": 13.057930483419897,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 32684
    },
    {
      "epoch": 13.058330003995206,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 32685
    },
    {
      "epoch": 13.058729524570515,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 32686
    },
    {
      "epoch": 13.059129045145825,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 32687
    },
    {
      "epoch": 13.059528565721134,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 32688
    },
    {
      "epoch": 13.059928086296445,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 32689
    },
    {
      "epoch": 13.060327606871754,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 32690
    },
    {
      "epoch": 13.060727127447063,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 32691
    },
    {
      "epoch": 13.061126648022373,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 32692
    },
    {
      "epoch": 13.061526168597682,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 32693
    },
    {
      "epoch": 13.061925689172993,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 32694
    },
    {
      "epoch": 13.062325209748302,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 32695
    },
    {
      "epoch": 13.062724730323612,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 32696
    },
    {
      "epoch": 13.063124250898921,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 32697
    },
    {
      "epoch": 13.06352377147423,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 32698
    },
    {
      "epoch": 13.063923292049541,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 32699
    },
    {
      "epoch": 13.06432281262485,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 32700
    },
    {
      "epoch": 13.06472233320016,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 32701
    },
    {
      "epoch": 13.06512185377547,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 32702
    },
    {
      "epoch": 13.065521374350778,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0128,
      "step": 32703
    },
    {
      "epoch": 13.065920894926089,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 32704
    },
    {
      "epoch": 13.066320415501398,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 32705
    },
    {
      "epoch": 13.066719936076709,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 32706
    },
    {
      "epoch": 13.067119456652017,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 32707
    },
    {
      "epoch": 13.067518977227326,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 32708
    },
    {
      "epoch": 13.067918497802637,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 32709
    },
    {
      "epoch": 13.068318018377946,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 32710
    },
    {
      "epoch": 13.068717538953257,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 32711
    },
    {
      "epoch": 13.069117059528566,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 32712
    },
    {
      "epoch": 13.069516580103876,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 32713
    },
    {
      "epoch": 13.069916100679185,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 32714
    },
    {
      "epoch": 13.070315621254494,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 32715
    },
    {
      "epoch": 13.070715141829805,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 32716
    },
    {
      "epoch": 13.071114662405114,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 32717
    },
    {
      "epoch": 13.071514182980424,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 32718
    },
    {
      "epoch": 13.071913703555733,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 32719
    },
    {
      "epoch": 13.072313224131042,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 32720
    },
    {
      "epoch": 13.072712744706353,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 32721
    },
    {
      "epoch": 13.073112265281662,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 32722
    },
    {
      "epoch": 13.073511785856972,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 32723
    },
    {
      "epoch": 13.073911306432281,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0458,
      "step": 32724
    },
    {
      "epoch": 13.07431082700759,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 32725
    },
    {
      "epoch": 13.0747103475829,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 32726
    },
    {
      "epoch": 13.07510986815821,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 32727
    },
    {
      "epoch": 13.07550938873352,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 32728
    },
    {
      "epoch": 13.07590890930883,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 32729
    },
    {
      "epoch": 13.07630842988414,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 32730
    },
    {
      "epoch": 13.076707950459449,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 32731
    },
    {
      "epoch": 13.077107471034758,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 32732
    },
    {
      "epoch": 13.077506991610068,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 32733
    },
    {
      "epoch": 13.077906512185377,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 32734
    },
    {
      "epoch": 13.078306032760688,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 32735
    },
    {
      "epoch": 13.078705553335997,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 32736
    },
    {
      "epoch": 13.079105073911306,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 32737
    },
    {
      "epoch": 13.079504594486616,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 32738
    },
    {
      "epoch": 13.079904115061925,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 32739
    },
    {
      "epoch": 13.080303635637236,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 32740
    },
    {
      "epoch": 13.080703156212545,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 32741
    },
    {
      "epoch": 13.081102676787854,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 32742
    },
    {
      "epoch": 13.081502197363164,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 32743
    },
    {
      "epoch": 13.081901717938473,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 32744
    },
    {
      "epoch": 13.082301238513784,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 32745
    },
    {
      "epoch": 13.082700759089093,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 32746
    },
    {
      "epoch": 13.083100279664404,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 32747
    },
    {
      "epoch": 13.083499800239712,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 32748
    },
    {
      "epoch": 13.083899320815021,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 32749
    },
    {
      "epoch": 13.084298841390332,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 32750
    },
    {
      "epoch": 13.084698361965641,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 32751
    },
    {
      "epoch": 13.085097882540952,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 32752
    },
    {
      "epoch": 13.08549740311626,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 32753
    },
    {
      "epoch": 13.08589692369157,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 32754
    },
    {
      "epoch": 13.08629644426688,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 32755
    },
    {
      "epoch": 13.086695964842189,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 32756
    },
    {
      "epoch": 13.0870954854175,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0036,
      "step": 32757
    },
    {
      "epoch": 13.087495005992809,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 32758
    },
    {
      "epoch": 13.087894526568117,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 32759
    },
    {
      "epoch": 13.088294047143428,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 32760
    },
    {
      "epoch": 13.088693567718737,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 32761
    },
    {
      "epoch": 13.089093088294048,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 32762
    },
    {
      "epoch": 13.089492608869357,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 32763
    },
    {
      "epoch": 13.089892129444667,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 32764
    },
    {
      "epoch": 13.090291650019976,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 32765
    },
    {
      "epoch": 13.090691170595285,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 32766
    },
    {
      "epoch": 13.091090691170596,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 32767
    },
    {
      "epoch": 13.091490211745905,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 32768
    },
    {
      "epoch": 13.091889732321215,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 32769
    },
    {
      "epoch": 13.092289252896524,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 32770
    },
    {
      "epoch": 13.092688773471833,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 32771
    },
    {
      "epoch": 13.093088294047144,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 32772
    },
    {
      "epoch": 13.093487814622453,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 32773
    },
    {
      "epoch": 13.093887335197763,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 32774
    },
    {
      "epoch": 13.094286855773072,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 32775
    },
    {
      "epoch": 13.094686376348381,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 32776
    },
    {
      "epoch": 13.095085896923692,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 32777
    },
    {
      "epoch": 13.095485417499,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 32778
    },
    {
      "epoch": 13.095884938074311,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 32779
    },
    {
      "epoch": 13.09628445864962,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 32780
    },
    {
      "epoch": 13.09668397922493,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 32781
    },
    {
      "epoch": 13.09708349980024,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 32782
    },
    {
      "epoch": 13.097483020375549,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 32783
    },
    {
      "epoch": 13.09788254095086,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 32784
    },
    {
      "epoch": 13.098282061526168,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 32785
    },
    {
      "epoch": 13.098681582101479,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 32786
    },
    {
      "epoch": 13.099081102676788,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 32787
    },
    {
      "epoch": 13.099480623252097,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 32788
    },
    {
      "epoch": 13.099880143827408,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 32789
    },
    {
      "epoch": 13.100279664402716,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 32790
    },
    {
      "epoch": 13.100679184978027,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 32791
    },
    {
      "epoch": 13.101078705553336,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 32792
    },
    {
      "epoch": 13.101478226128645,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 32793
    },
    {
      "epoch": 13.101877746703956,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 32794
    },
    {
      "epoch": 13.102277267279264,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 32795
    },
    {
      "epoch": 13.102676787854575,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 32796
    },
    {
      "epoch": 13.103076308429884,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 32797
    },
    {
      "epoch": 13.103475829005193,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 32798
    },
    {
      "epoch": 13.103875349580504,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 32799
    },
    {
      "epoch": 13.104274870155812,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 32800
    },
    {
      "epoch": 13.104674390731123,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 32801
    },
    {
      "epoch": 13.105073911306432,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 32802
    },
    {
      "epoch": 13.105473431881743,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 32803
    },
    {
      "epoch": 13.105872952457052,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 32804
    },
    {
      "epoch": 13.10627247303236,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 32805
    },
    {
      "epoch": 13.106671993607671,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 32806
    },
    {
      "epoch": 13.10707151418298,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 32807
    },
    {
      "epoch": 13.10747103475829,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 32808
    },
    {
      "epoch": 13.1078705553336,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 32809
    },
    {
      "epoch": 13.108270075908909,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 32810
    },
    {
      "epoch": 13.10866959648422,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 32811
    },
    {
      "epoch": 13.109069117059528,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 32812
    },
    {
      "epoch": 13.109468637634839,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 32813
    },
    {
      "epoch": 13.109868158210148,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 32814
    },
    {
      "epoch": 13.110267678785457,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 32815
    },
    {
      "epoch": 13.110667199360767,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 32816
    },
    {
      "epoch": 13.111066719936076,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 32817
    },
    {
      "epoch": 13.111466240511387,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 32818
    },
    {
      "epoch": 13.111865761086696,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 32819
    },
    {
      "epoch": 13.112265281662006,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 32820
    },
    {
      "epoch": 13.112664802237315,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 32821
    },
    {
      "epoch": 13.113064322812624,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 32822
    },
    {
      "epoch": 13.113463843387935,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 32823
    },
    {
      "epoch": 13.113863363963244,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 32824
    },
    {
      "epoch": 13.114262884538554,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 32825
    },
    {
      "epoch": 13.114662405113863,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 32826
    },
    {
      "epoch": 13.115061925689172,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 32827
    },
    {
      "epoch": 13.115461446264483,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 32828
    },
    {
      "epoch": 13.115860966839792,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 32829
    },
    {
      "epoch": 13.116260487415103,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 32830
    },
    {
      "epoch": 13.116660007990411,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 32831
    },
    {
      "epoch": 13.11705952856572,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 32832
    },
    {
      "epoch": 13.117459049141031,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 32833
    },
    {
      "epoch": 13.11785856971634,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 32834
    },
    {
      "epoch": 13.11825809029165,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 32835
    },
    {
      "epoch": 13.11865761086696,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 32836
    },
    {
      "epoch": 13.11905713144227,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 32837
    },
    {
      "epoch": 13.119456652017579,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 32838
    },
    {
      "epoch": 13.119856172592888,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 32839
    },
    {
      "epoch": 13.120255693168199,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 32840
    },
    {
      "epoch": 13.120655213743508,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 32841
    },
    {
      "epoch": 13.121054734318818,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 32842
    },
    {
      "epoch": 13.121454254894127,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 32843
    },
    {
      "epoch": 13.121853775469436,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 32844
    },
    {
      "epoch": 13.122253296044747,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 32845
    },
    {
      "epoch": 13.122652816620056,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 32846
    },
    {
      "epoch": 13.123052337195366,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 32847
    },
    {
      "epoch": 13.123451857770675,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 32848
    },
    {
      "epoch": 13.123851378345984,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 32849
    },
    {
      "epoch": 13.124250898921295,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 32850
    },
    {
      "epoch": 13.124650419496604,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 32851
    },
    {
      "epoch": 13.125049940071914,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 32852
    },
    {
      "epoch": 13.125449460647223,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 32853
    },
    {
      "epoch": 13.125848981222532,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 32854
    },
    {
      "epoch": 13.126248501797843,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 32855
    },
    {
      "epoch": 13.126648022373152,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 32856
    },
    {
      "epoch": 13.127047542948462,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 32857
    },
    {
      "epoch": 13.127447063523771,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 32858
    },
    {
      "epoch": 13.127846584099082,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 32859
    },
    {
      "epoch": 13.12824610467439,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 32860
    },
    {
      "epoch": 13.1286456252497,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0121,
      "step": 32861
    },
    {
      "epoch": 13.12904514582501,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 32862
    },
    {
      "epoch": 13.12944466640032,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 32863
    },
    {
      "epoch": 13.12984418697563,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 32864
    },
    {
      "epoch": 13.130243707550939,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 32865
    },
    {
      "epoch": 13.130643228126248,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 32866
    },
    {
      "epoch": 13.131042748701558,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 32867
    },
    {
      "epoch": 13.131442269276867,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 32868
    },
    {
      "epoch": 13.131841789852178,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 32869
    },
    {
      "epoch": 13.132241310427487,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 32870
    },
    {
      "epoch": 13.132640831002796,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 32871
    },
    {
      "epoch": 13.133040351578106,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 32872
    },
    {
      "epoch": 13.133439872153415,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 32873
    },
    {
      "epoch": 13.133839392728726,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 32874
    },
    {
      "epoch": 13.134238913304035,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 32875
    },
    {
      "epoch": 13.134638433879346,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 32876
    },
    {
      "epoch": 13.135037954454654,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 32877
    },
    {
      "epoch": 13.135437475029963,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 32878
    },
    {
      "epoch": 13.135836995605274,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 32879
    },
    {
      "epoch": 13.136236516180583,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 32880
    },
    {
      "epoch": 13.136636036755894,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 32881
    },
    {
      "epoch": 13.137035557331203,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 32882
    },
    {
      "epoch": 13.137435077906511,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 32883
    },
    {
      "epoch": 13.137834598481822,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 32884
    },
    {
      "epoch": 13.138234119057131,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 32885
    },
    {
      "epoch": 13.138633639632442,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 32886
    },
    {
      "epoch": 13.13903316020775,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 32887
    },
    {
      "epoch": 13.13943268078306,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 32888
    },
    {
      "epoch": 13.13983220135837,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 32889
    },
    {
      "epoch": 13.140231721933679,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 32890
    },
    {
      "epoch": 13.14063124250899,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 32891
    },
    {
      "epoch": 13.141030763084299,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 32892
    },
    {
      "epoch": 13.14143028365961,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 32893
    },
    {
      "epoch": 13.141829804234918,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 32894
    },
    {
      "epoch": 13.142229324810227,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 32895
    },
    {
      "epoch": 13.142628845385538,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 32896
    },
    {
      "epoch": 13.143028365960847,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 32897
    },
    {
      "epoch": 13.143427886536157,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 32898
    },
    {
      "epoch": 13.143827407111466,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 32899
    },
    {
      "epoch": 13.144226927686775,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 32900
    },
    {
      "epoch": 13.144626448262086,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 32901
    },
    {
      "epoch": 13.145025968837395,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 32902
    },
    {
      "epoch": 13.145425489412705,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 32903
    },
    {
      "epoch": 13.145825009988014,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 32904
    },
    {
      "epoch": 13.146224530563323,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 32905
    },
    {
      "epoch": 13.146624051138634,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 32906
    },
    {
      "epoch": 13.147023571713943,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 32907
    },
    {
      "epoch": 13.147423092289253,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 32908
    },
    {
      "epoch": 13.147822612864562,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 32909
    },
    {
      "epoch": 13.148222133439873,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 32910
    },
    {
      "epoch": 13.148621654015182,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 32911
    },
    {
      "epoch": 13.14902117459049,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 32912
    },
    {
      "epoch": 13.149420695165801,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 32913
    },
    {
      "epoch": 13.14982021574111,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 32914
    },
    {
      "epoch": 13.150219736316421,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 32915
    },
    {
      "epoch": 13.15061925689173,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 32916
    },
    {
      "epoch": 13.151018777467039,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 32917
    },
    {
      "epoch": 13.15141829804235,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 32918
    },
    {
      "epoch": 13.151817818617658,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 32919
    },
    {
      "epoch": 13.152217339192969,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 32920
    },
    {
      "epoch": 13.152616859768278,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 32921
    },
    {
      "epoch": 13.153016380343587,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 32922
    },
    {
      "epoch": 13.153415900918898,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 32923
    },
    {
      "epoch": 13.153815421494206,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 32924
    },
    {
      "epoch": 13.154214942069517,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 32925
    },
    {
      "epoch": 13.154614462644826,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 32926
    },
    {
      "epoch": 13.155013983220137,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 32927
    },
    {
      "epoch": 13.155413503795446,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 32928
    },
    {
      "epoch": 13.155813024370755,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 32929
    },
    {
      "epoch": 13.156212544946065,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 32930
    },
    {
      "epoch": 13.156612065521374,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 32931
    },
    {
      "epoch": 13.157011586096685,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 32932
    },
    {
      "epoch": 13.157411106671994,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 32933
    },
    {
      "epoch": 13.157810627247303,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 32934
    },
    {
      "epoch": 13.158210147822613,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 32935
    },
    {
      "epoch": 13.158609668397922,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 32936
    },
    {
      "epoch": 13.159009188973233,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 32937
    },
    {
      "epoch": 13.159408709548542,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 32938
    },
    {
      "epoch": 13.15980823012385,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 32939
    },
    {
      "epoch": 13.160207750699161,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 32940
    },
    {
      "epoch": 13.16060727127447,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 32941
    },
    {
      "epoch": 13.16100679184978,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 32942
    },
    {
      "epoch": 13.16140631242509,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 32943
    },
    {
      "epoch": 13.1618058330004,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 32944
    },
    {
      "epoch": 13.16220535357571,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 32945
    },
    {
      "epoch": 13.162604874151018,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 32946
    },
    {
      "epoch": 13.163004394726329,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 32947
    },
    {
      "epoch": 13.163403915301638,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 32948
    },
    {
      "epoch": 13.163803435876948,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 32949
    },
    {
      "epoch": 13.164202956452257,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 32950
    },
    {
      "epoch": 13.164602477027566,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 32951
    },
    {
      "epoch": 13.165001997602877,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 32952
    },
    {
      "epoch": 13.165401518178186,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 32953
    },
    {
      "epoch": 13.165801038753496,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 32954
    },
    {
      "epoch": 13.166200559328805,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 32955
    },
    {
      "epoch": 13.166600079904114,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 32956
    },
    {
      "epoch": 13.166999600479425,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 32957
    },
    {
      "epoch": 13.167399121054734,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 32958
    },
    {
      "epoch": 13.167798641630045,
      "grad_norm": 0.171875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 32959
    },
    {
      "epoch": 13.168198162205353,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 32960
    },
    {
      "epoch": 13.168597682780662,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 32961
    },
    {
      "epoch": 13.168997203355973,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 32962
    },
    {
      "epoch": 13.169396723931282,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 32963
    },
    {
      "epoch": 13.169796244506593,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 32964
    },
    {
      "epoch": 13.170195765081901,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 32965
    },
    {
      "epoch": 13.170595285657212,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 32966
    },
    {
      "epoch": 13.170994806232521,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 32967
    },
    {
      "epoch": 13.17139432680783,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 32968
    },
    {
      "epoch": 13.17179384738314,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 32969
    },
    {
      "epoch": 13.17219336795845,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 32970
    },
    {
      "epoch": 13.17259288853376,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 32971
    },
    {
      "epoch": 13.172992409109069,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 32972
    },
    {
      "epoch": 13.173391929684378,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 32973
    },
    {
      "epoch": 13.173791450259689,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 32974
    },
    {
      "epoch": 13.174190970834998,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 32975
    },
    {
      "epoch": 13.174590491410308,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 32976
    },
    {
      "epoch": 13.174990011985617,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 32977
    },
    {
      "epoch": 13.175389532560926,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 32978
    },
    {
      "epoch": 13.175789053136237,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 32979
    },
    {
      "epoch": 13.176188573711546,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 32980
    },
    {
      "epoch": 13.176588094286856,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 32981
    },
    {
      "epoch": 13.176987614862165,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 32982
    },
    {
      "epoch": 13.177387135437476,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 32983
    },
    {
      "epoch": 13.177786656012785,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 32984
    },
    {
      "epoch": 13.178186176588094,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 32985
    },
    {
      "epoch": 13.178585697163404,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 32986
    },
    {
      "epoch": 13.178985217738713,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 32987
    },
    {
      "epoch": 13.179384738314024,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 32988
    },
    {
      "epoch": 13.179784258889333,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 32989
    },
    {
      "epoch": 13.180183779464642,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 32990
    },
    {
      "epoch": 13.180583300039952,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 32991
    },
    {
      "epoch": 13.180982820615261,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 32992
    },
    {
      "epoch": 13.181382341190572,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 32993
    },
    {
      "epoch": 13.18178186176588,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 32994
    },
    {
      "epoch": 13.18218138234119,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 32995
    },
    {
      "epoch": 13.1825809029165,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 32996
    },
    {
      "epoch": 13.18298042349181,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 32997
    },
    {
      "epoch": 13.18337994406712,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 32998
    },
    {
      "epoch": 13.183779464642429,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 32999
    },
    {
      "epoch": 13.18417898521774,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 33000
    },
    {
      "epoch": 13.184578505793048,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 33001
    },
    {
      "epoch": 13.184978026368357,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 33002
    },
    {
      "epoch": 13.185377546943668,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 33003
    },
    {
      "epoch": 13.185777067518977,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 33004
    },
    {
      "epoch": 13.186176588094288,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 33005
    },
    {
      "epoch": 13.186576108669597,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 33006
    },
    {
      "epoch": 13.186975629244905,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 33007
    },
    {
      "epoch": 13.187375149820216,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 33008
    },
    {
      "epoch": 13.187774670395525,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 33009
    },
    {
      "epoch": 13.188174190970836,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 33010
    },
    {
      "epoch": 13.188573711546145,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 33011
    },
    {
      "epoch": 13.188973232121453,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 33012
    },
    {
      "epoch": 13.189372752696764,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 33013
    },
    {
      "epoch": 13.189772273272073,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 33014
    },
    {
      "epoch": 13.190171793847384,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 33015
    },
    {
      "epoch": 13.190571314422693,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 33016
    },
    {
      "epoch": 13.190970834998003,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 33017
    },
    {
      "epoch": 13.191370355573312,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 33018
    },
    {
      "epoch": 13.191769876148621,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 33019
    },
    {
      "epoch": 13.192169396723932,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 33020
    },
    {
      "epoch": 13.19256891729924,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0075,
      "step": 33021
    },
    {
      "epoch": 13.192968437874551,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 33022
    },
    {
      "epoch": 13.19336795844986,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 33023
    },
    {
      "epoch": 13.193767479025169,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 33024
    },
    {
      "epoch": 13.19416699960048,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 33025
    },
    {
      "epoch": 13.194566520175789,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 33026
    },
    {
      "epoch": 13.1949660407511,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 33027
    },
    {
      "epoch": 13.195365561326408,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 33028
    },
    {
      "epoch": 13.195765081901717,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 33029
    },
    {
      "epoch": 13.196164602477028,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 33030
    },
    {
      "epoch": 13.196564123052337,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 33031
    },
    {
      "epoch": 13.196963643627647,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 33032
    },
    {
      "epoch": 13.197363164202956,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 33033
    },
    {
      "epoch": 13.197762684778265,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 33034
    },
    {
      "epoch": 13.198162205353576,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 33035
    },
    {
      "epoch": 13.198561725928885,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 33036
    },
    {
      "epoch": 13.198961246504195,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 33037
    },
    {
      "epoch": 13.199360767079504,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 33038
    },
    {
      "epoch": 13.199760287654815,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 33039
    },
    {
      "epoch": 13.200159808230124,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 33040
    },
    {
      "epoch": 13.200559328805433,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 33041
    },
    {
      "epoch": 13.200958849380743,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 33042
    },
    {
      "epoch": 13.201358369956052,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 33043
    },
    {
      "epoch": 13.201757890531363,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 33044
    },
    {
      "epoch": 13.202157411106672,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 33045
    },
    {
      "epoch": 13.20255693168198,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 33046
    },
    {
      "epoch": 13.202956452257292,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 33047
    },
    {
      "epoch": 13.2033559728326,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 33048
    },
    {
      "epoch": 13.203755493407911,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 33049
    },
    {
      "epoch": 13.20415501398322,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 33050
    },
    {
      "epoch": 13.204554534558529,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 33051
    },
    {
      "epoch": 13.20495405513384,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 33052
    },
    {
      "epoch": 13.205353575709148,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 33053
    },
    {
      "epoch": 13.20575309628446,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 33054
    },
    {
      "epoch": 13.206152616859768,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 33055
    },
    {
      "epoch": 13.206552137435079,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 33056
    },
    {
      "epoch": 13.206951658010388,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 33057
    },
    {
      "epoch": 13.207351178585697,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 33058
    },
    {
      "epoch": 13.207750699161007,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 33059
    },
    {
      "epoch": 13.208150219736316,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 33060
    },
    {
      "epoch": 13.208549740311627,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 33061
    },
    {
      "epoch": 13.208949260886936,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 33062
    },
    {
      "epoch": 13.209348781462245,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 33063
    },
    {
      "epoch": 13.209748302037555,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0525,
      "step": 33064
    },
    {
      "epoch": 13.210147822612864,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 33065
    },
    {
      "epoch": 13.210547343188175,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 33066
    },
    {
      "epoch": 13.210946863763484,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 33067
    },
    {
      "epoch": 13.211346384338793,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 33068
    },
    {
      "epoch": 13.211745904914103,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 33069
    },
    {
      "epoch": 13.212145425489412,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 33070
    },
    {
      "epoch": 13.212544946064723,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 33071
    },
    {
      "epoch": 13.212944466640032,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 33072
    },
    {
      "epoch": 13.213343987215342,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 33073
    },
    {
      "epoch": 13.213743507790651,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 33074
    },
    {
      "epoch": 13.21414302836596,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 33075
    },
    {
      "epoch": 13.214542548941271,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 33076
    },
    {
      "epoch": 13.21494206951658,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 33077
    },
    {
      "epoch": 13.21534159009189,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0117,
      "step": 33078
    },
    {
      "epoch": 13.2157411106672,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 33079
    },
    {
      "epoch": 13.216140631242508,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 33080
    },
    {
      "epoch": 13.216540151817819,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 33081
    },
    {
      "epoch": 13.216939672393128,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 33082
    },
    {
      "epoch": 13.217339192968439,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 33083
    },
    {
      "epoch": 13.217738713543747,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 33084
    },
    {
      "epoch": 13.218138234119056,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 33085
    },
    {
      "epoch": 13.218537754694367,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 33086
    },
    {
      "epoch": 13.218937275269676,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 33087
    },
    {
      "epoch": 13.219336795844987,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 33088
    },
    {
      "epoch": 13.219736316420295,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 33089
    },
    {
      "epoch": 13.220135836995606,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 33090
    },
    {
      "epoch": 13.220535357570915,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 33091
    },
    {
      "epoch": 13.220934878146224,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 33092
    },
    {
      "epoch": 13.221334398721535,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 33093
    },
    {
      "epoch": 13.221733919296843,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 33094
    },
    {
      "epoch": 13.222133439872154,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 33095
    },
    {
      "epoch": 13.222532960447463,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 33096
    },
    {
      "epoch": 13.222932481022772,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 33097
    },
    {
      "epoch": 13.223332001598083,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 33098
    },
    {
      "epoch": 13.223731522173392,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 33099
    },
    {
      "epoch": 13.224131042748702,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 33100
    },
    {
      "epoch": 13.224530563324011,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 33101
    },
    {
      "epoch": 13.22493008389932,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 33102
    },
    {
      "epoch": 13.22532960447463,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 33103
    },
    {
      "epoch": 13.22572912504994,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 33104
    },
    {
      "epoch": 13.22612864562525,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 33105
    },
    {
      "epoch": 13.22652816620056,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 33106
    },
    {
      "epoch": 13.22692768677587,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 33107
    },
    {
      "epoch": 13.227327207351179,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 33108
    },
    {
      "epoch": 13.227726727926488,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 33109
    },
    {
      "epoch": 13.228126248501798,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 33110
    },
    {
      "epoch": 13.228525769077107,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 33111
    },
    {
      "epoch": 13.228925289652418,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 33112
    },
    {
      "epoch": 13.229324810227727,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 33113
    },
    {
      "epoch": 13.229724330803036,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 33114
    },
    {
      "epoch": 13.230123851378346,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 33115
    },
    {
      "epoch": 13.230523371953655,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 33116
    },
    {
      "epoch": 13.230922892528966,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 33117
    },
    {
      "epoch": 13.231322413104275,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 33118
    },
    {
      "epoch": 13.231721933679584,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 33119
    },
    {
      "epoch": 13.232121454254894,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 33120
    },
    {
      "epoch": 13.232520974830203,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 33121
    },
    {
      "epoch": 13.232920495405514,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 33122
    },
    {
      "epoch": 13.233320015980823,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 33123
    },
    {
      "epoch": 13.233719536556134,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 33124
    },
    {
      "epoch": 13.234119057131442,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 33125
    },
    {
      "epoch": 13.234518577706751,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 33126
    },
    {
      "epoch": 13.234918098282062,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 33127
    },
    {
      "epoch": 13.235317618857371,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 33128
    },
    {
      "epoch": 13.235717139432682,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 33129
    },
    {
      "epoch": 13.23611666000799,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 33130
    },
    {
      "epoch": 13.2365161805833,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 33131
    },
    {
      "epoch": 13.23691570115861,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 33132
    },
    {
      "epoch": 13.237315221733919,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 33133
    },
    {
      "epoch": 13.23771474230923,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 33134
    },
    {
      "epoch": 13.238114262884539,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 33135
    },
    {
      "epoch": 13.238513783459847,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 33136
    },
    {
      "epoch": 13.238913304035158,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 33137
    },
    {
      "epoch": 13.239312824610467,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 33138
    },
    {
      "epoch": 13.239712345185778,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 33139
    },
    {
      "epoch": 13.240111865761087,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 33140
    },
    {
      "epoch": 13.240511386336395,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 33141
    },
    {
      "epoch": 13.240910906911706,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 33142
    },
    {
      "epoch": 13.241310427487015,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 33143
    },
    {
      "epoch": 13.241709948062326,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 33144
    },
    {
      "epoch": 13.242109468637635,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 33145
    },
    {
      "epoch": 13.242508989212945,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 33146
    },
    {
      "epoch": 13.242908509788254,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 33147
    },
    {
      "epoch": 13.243308030363563,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 33148
    },
    {
      "epoch": 13.243707550938874,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 33149
    },
    {
      "epoch": 13.244107071514183,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 33150
    },
    {
      "epoch": 13.244506592089493,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 33151
    },
    {
      "epoch": 13.244906112664802,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 33152
    },
    {
      "epoch": 13.245305633240111,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 33153
    },
    {
      "epoch": 13.245705153815422,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 33154
    },
    {
      "epoch": 13.24610467439073,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 33155
    },
    {
      "epoch": 13.246504194966041,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 33156
    },
    {
      "epoch": 13.24690371554135,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 33157
    },
    {
      "epoch": 13.24730323611666,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 33158
    },
    {
      "epoch": 13.24770275669197,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 33159
    },
    {
      "epoch": 13.248102277267279,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 33160
    },
    {
      "epoch": 13.24850179784259,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 33161
    },
    {
      "epoch": 13.248901318417898,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 33162
    },
    {
      "epoch": 13.249300838993209,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 33163
    },
    {
      "epoch": 13.249700359568518,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 33164
    },
    {
      "epoch": 13.250099880143827,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 33165
    },
    {
      "epoch": 13.250499400719137,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 33166
    },
    {
      "epoch": 13.250898921294446,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 33167
    },
    {
      "epoch": 13.251298441869757,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 33168
    },
    {
      "epoch": 13.251697962445066,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 33169
    },
    {
      "epoch": 13.252097483020375,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 33170
    },
    {
      "epoch": 13.252497003595685,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 33171
    },
    {
      "epoch": 13.252896524170994,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 33172
    },
    {
      "epoch": 13.253296044746305,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 33173
    },
    {
      "epoch": 13.253695565321614,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 33174
    },
    {
      "epoch": 13.254095085896923,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 33175
    },
    {
      "epoch": 13.254494606472234,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 33176
    },
    {
      "epoch": 13.254894127047542,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 33177
    },
    {
      "epoch": 13.255293647622853,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 33178
    },
    {
      "epoch": 13.255693168198162,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 33179
    },
    {
      "epoch": 13.256092688773473,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 33180
    },
    {
      "epoch": 13.256492209348782,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 33181
    },
    {
      "epoch": 13.25689172992409,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 33182
    },
    {
      "epoch": 13.257291250499401,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 33183
    },
    {
      "epoch": 13.25769077107471,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 33184
    },
    {
      "epoch": 13.25809029165002,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 33185
    },
    {
      "epoch": 13.25848981222533,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 33186
    },
    {
      "epoch": 13.258889332800639,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 33187
    },
    {
      "epoch": 13.25928885337595,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 33188
    },
    {
      "epoch": 13.259688373951258,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 33189
    },
    {
      "epoch": 13.260087894526569,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 33190
    },
    {
      "epoch": 13.260487415101878,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 33191
    },
    {
      "epoch": 13.260886935677187,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 33192
    },
    {
      "epoch": 13.261286456252497,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 33193
    },
    {
      "epoch": 13.261685976827806,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 33194
    },
    {
      "epoch": 13.262085497403117,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 33195
    },
    {
      "epoch": 13.262485017978426,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 33196
    },
    {
      "epoch": 13.262884538553735,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 33197
    },
    {
      "epoch": 13.263284059129045,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 33198
    },
    {
      "epoch": 13.263683579704354,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 33199
    },
    {
      "epoch": 13.264083100279665,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 33200
    },
    {
      "epoch": 13.264482620854974,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 33201
    },
    {
      "epoch": 13.264882141430284,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 33202
    },
    {
      "epoch": 13.265281662005593,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 33203
    },
    {
      "epoch": 13.265681182580902,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 33204
    },
    {
      "epoch": 13.266080703156213,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 33205
    },
    {
      "epoch": 13.266480223731522,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 33206
    },
    {
      "epoch": 13.266879744306832,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 33207
    },
    {
      "epoch": 13.267279264882141,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 33208
    },
    {
      "epoch": 13.26767878545745,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 33209
    },
    {
      "epoch": 13.268078306032761,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 33210
    },
    {
      "epoch": 13.26847782660807,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 33211
    },
    {
      "epoch": 13.26887734718338,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 33212
    },
    {
      "epoch": 13.26927686775869,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 33213
    },
    {
      "epoch": 13.269676388333998,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 33214
    },
    {
      "epoch": 13.270075908909309,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 33215
    },
    {
      "epoch": 13.270475429484618,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 33216
    },
    {
      "epoch": 13.270874950059929,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 33217
    },
    {
      "epoch": 13.271274470635237,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 33218
    },
    {
      "epoch": 13.271673991210548,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 33219
    },
    {
      "epoch": 13.272073511785857,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 33220
    },
    {
      "epoch": 13.272473032361166,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 33221
    },
    {
      "epoch": 13.272872552936477,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 33222
    },
    {
      "epoch": 13.273272073511786,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 33223
    },
    {
      "epoch": 13.273671594087096,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 33224
    },
    {
      "epoch": 13.274071114662405,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 33225
    },
    {
      "epoch": 13.274470635237714,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 33226
    },
    {
      "epoch": 13.274870155813025,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 33227
    },
    {
      "epoch": 13.275269676388334,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 33228
    },
    {
      "epoch": 13.275669196963644,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 33229
    },
    {
      "epoch": 13.276068717538953,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 33230
    },
    {
      "epoch": 13.276468238114262,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0126,
      "step": 33231
    },
    {
      "epoch": 13.276867758689573,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 33232
    },
    {
      "epoch": 13.277267279264882,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 33233
    },
    {
      "epoch": 13.277666799840192,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 33234
    },
    {
      "epoch": 13.278066320415501,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 33235
    },
    {
      "epoch": 13.278465840990812,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 33236
    },
    {
      "epoch": 13.27886536156612,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 33237
    },
    {
      "epoch": 13.27926488214143,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 33238
    },
    {
      "epoch": 13.27966440271674,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 33239
    },
    {
      "epoch": 13.28006392329205,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 33240
    },
    {
      "epoch": 13.28046344386736,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 33241
    },
    {
      "epoch": 13.280862964442669,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 33242
    },
    {
      "epoch": 13.281262485017978,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 33243
    },
    {
      "epoch": 13.281662005593288,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 33244
    },
    {
      "epoch": 13.282061526168597,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 33245
    },
    {
      "epoch": 13.282461046743908,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 33246
    },
    {
      "epoch": 13.282860567319217,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 33247
    },
    {
      "epoch": 13.283260087894526,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 33248
    },
    {
      "epoch": 13.283659608469836,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 33249
    },
    {
      "epoch": 13.284059129045145,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 33250
    },
    {
      "epoch": 13.284458649620456,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 33251
    },
    {
      "epoch": 13.284858170195765,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 33252
    },
    {
      "epoch": 13.285257690771076,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 33253
    },
    {
      "epoch": 13.285657211346384,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 33254
    },
    {
      "epoch": 13.286056731921693,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 33255
    },
    {
      "epoch": 13.286456252497004,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 33256
    },
    {
      "epoch": 13.286855773072313,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 33257
    },
    {
      "epoch": 13.287255293647624,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 33258
    },
    {
      "epoch": 13.287654814222932,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 33259
    },
    {
      "epoch": 13.288054334798241,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 33260
    },
    {
      "epoch": 13.288453855373552,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 33261
    },
    {
      "epoch": 13.288853375948861,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 33262
    },
    {
      "epoch": 13.289252896524172,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 33263
    },
    {
      "epoch": 13.28965241709948,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 33264
    },
    {
      "epoch": 13.29005193767479,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 33265
    },
    {
      "epoch": 13.2904514582501,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 33266
    },
    {
      "epoch": 13.290850978825409,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 33267
    },
    {
      "epoch": 13.29125049940072,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 33268
    },
    {
      "epoch": 13.291650019976029,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 33269
    },
    {
      "epoch": 13.29204954055134,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 33270
    },
    {
      "epoch": 13.292449061126648,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 33271
    },
    {
      "epoch": 13.292848581701957,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 33272
    },
    {
      "epoch": 13.293248102277268,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 33273
    },
    {
      "epoch": 13.293647622852577,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 33274
    },
    {
      "epoch": 13.294047143427887,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 33275
    },
    {
      "epoch": 13.294446664003196,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 33276
    },
    {
      "epoch": 13.294846184578505,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 33277
    },
    {
      "epoch": 13.295245705153816,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 33278
    },
    {
      "epoch": 13.295645225729125,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 33279
    },
    {
      "epoch": 13.296044746304435,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 33280
    },
    {
      "epoch": 13.296444266879744,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 33281
    },
    {
      "epoch": 13.296843787455053,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 33282
    },
    {
      "epoch": 13.297243308030364,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 33283
    },
    {
      "epoch": 13.297642828605673,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 33284
    },
    {
      "epoch": 13.298042349180983,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 33285
    },
    {
      "epoch": 13.298441869756292,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 33286
    },
    {
      "epoch": 13.298841390331603,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 33287
    },
    {
      "epoch": 13.299240910906912,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 33288
    },
    {
      "epoch": 13.29964043148222,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 33289
    },
    {
      "epoch": 13.300039952057531,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 33290
    },
    {
      "epoch": 13.30043947263284,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 33291
    },
    {
      "epoch": 13.300838993208151,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 33292
    },
    {
      "epoch": 13.30123851378346,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 33293
    },
    {
      "epoch": 13.301638034358769,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 33294
    },
    {
      "epoch": 13.30203755493408,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 33295
    },
    {
      "epoch": 13.302437075509388,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 33296
    },
    {
      "epoch": 13.302836596084699,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 33297
    },
    {
      "epoch": 13.303236116660008,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 33298
    },
    {
      "epoch": 13.303635637235317,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 33299
    },
    {
      "epoch": 13.304035157810628,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 33300
    },
    {
      "epoch": 13.304434678385936,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 33301
    },
    {
      "epoch": 13.304834198961247,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0468,
      "step": 33302
    },
    {
      "epoch": 13.305233719536556,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 33303
    },
    {
      "epoch": 13.305633240111867,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 33304
    },
    {
      "epoch": 13.306032760687176,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 33305
    },
    {
      "epoch": 13.306432281262484,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 33306
    },
    {
      "epoch": 13.306831801837795,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0135,
      "step": 33307
    },
    {
      "epoch": 13.307231322413104,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 33308
    },
    {
      "epoch": 13.307630842988415,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 33309
    },
    {
      "epoch": 13.308030363563724,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 33310
    },
    {
      "epoch": 13.308429884139032,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 33311
    },
    {
      "epoch": 13.308829404714343,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 33312
    },
    {
      "epoch": 13.309228925289652,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 33313
    },
    {
      "epoch": 13.309628445864963,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 33314
    },
    {
      "epoch": 13.310027966440272,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 33315
    },
    {
      "epoch": 13.31042748701558,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 33316
    },
    {
      "epoch": 13.310827007590891,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 33317
    },
    {
      "epoch": 13.3112265281662,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 33318
    },
    {
      "epoch": 13.31162604874151,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 33319
    },
    {
      "epoch": 13.31202556931682,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 33320
    },
    {
      "epoch": 13.31242508989213,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 33321
    },
    {
      "epoch": 13.31282461046744,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 33322
    },
    {
      "epoch": 13.313224131042748,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 33323
    },
    {
      "epoch": 13.313623651618059,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 33324
    },
    {
      "epoch": 13.314023172193368,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 33325
    },
    {
      "epoch": 13.314422692768678,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 33326
    },
    {
      "epoch": 13.314822213343987,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 33327
    },
    {
      "epoch": 13.315221733919296,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 33328
    },
    {
      "epoch": 13.315621254494607,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 33329
    },
    {
      "epoch": 13.316020775069916,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 33330
    },
    {
      "epoch": 13.316420295645226,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 33331
    },
    {
      "epoch": 13.316819816220535,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 33332
    },
    {
      "epoch": 13.317219336795844,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 33333
    },
    {
      "epoch": 13.317618857371155,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 33334
    },
    {
      "epoch": 13.318018377946464,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 33335
    },
    {
      "epoch": 13.318417898521774,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 33336
    },
    {
      "epoch": 13.318817419097083,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 33337
    },
    {
      "epoch": 13.319216939672392,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 33338
    },
    {
      "epoch": 13.319616460247703,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 33339
    },
    {
      "epoch": 13.320015980823012,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 33340
    },
    {
      "epoch": 13.320415501398323,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 33341
    },
    {
      "epoch": 13.320815021973631,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 33342
    },
    {
      "epoch": 13.321214542548942,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 33343
    },
    {
      "epoch": 13.321614063124251,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 33344
    },
    {
      "epoch": 13.32201358369956,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 33345
    },
    {
      "epoch": 13.32241310427487,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 33346
    },
    {
      "epoch": 13.32281262485018,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 33347
    },
    {
      "epoch": 13.32321214542549,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 33348
    },
    {
      "epoch": 13.323611666000799,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 33349
    },
    {
      "epoch": 13.324011186576108,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 33350
    },
    {
      "epoch": 13.324410707151419,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 33351
    },
    {
      "epoch": 13.324810227726728,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 33352
    },
    {
      "epoch": 13.325209748302038,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 33353
    },
    {
      "epoch": 13.325609268877347,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 33354
    },
    {
      "epoch": 13.326008789452656,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 33355
    },
    {
      "epoch": 13.326408310027967,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 33356
    },
    {
      "epoch": 13.326807830603276,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 33357
    },
    {
      "epoch": 13.327207351178586,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 33358
    },
    {
      "epoch": 13.327606871753895,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 33359
    },
    {
      "epoch": 13.328006392329206,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 33360
    },
    {
      "epoch": 13.328405912904515,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 33361
    },
    {
      "epoch": 13.328805433479824,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 33362
    },
    {
      "epoch": 13.329204954055134,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 33363
    },
    {
      "epoch": 13.329604474630443,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 33364
    },
    {
      "epoch": 13.330003995205754,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 33365
    },
    {
      "epoch": 13.330403515781063,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 33366
    },
    {
      "epoch": 13.330803036356372,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 33367
    },
    {
      "epoch": 13.331202556931682,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 33368
    },
    {
      "epoch": 13.331602077506991,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 33369
    },
    {
      "epoch": 13.332001598082302,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 33370
    },
    {
      "epoch": 13.33240111865761,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 33371
    },
    {
      "epoch": 13.33280063923292,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 33372
    },
    {
      "epoch": 13.33320015980823,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 33373
    },
    {
      "epoch": 13.33359968038354,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 33374
    },
    {
      "epoch": 13.33399920095885,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 33375
    },
    {
      "epoch": 13.334398721534159,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 33376
    },
    {
      "epoch": 13.334798242109468,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 33377
    },
    {
      "epoch": 13.335197762684778,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 33378
    },
    {
      "epoch": 13.335597283260087,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 33379
    },
    {
      "epoch": 13.335996803835398,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 33380
    },
    {
      "epoch": 13.336396324410707,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 33381
    },
    {
      "epoch": 13.336795844986018,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 33382
    },
    {
      "epoch": 13.337195365561326,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 33383
    },
    {
      "epoch": 13.337594886136635,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 33384
    },
    {
      "epoch": 13.337994406711946,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 33385
    },
    {
      "epoch": 13.338393927287255,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 33386
    },
    {
      "epoch": 13.338793447862566,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 33387
    },
    {
      "epoch": 13.339192968437874,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 33388
    },
    {
      "epoch": 13.339592489013183,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 33389
    },
    {
      "epoch": 13.339992009588494,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 33390
    },
    {
      "epoch": 13.340391530163803,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 33391
    },
    {
      "epoch": 13.340791050739114,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 33392
    },
    {
      "epoch": 13.341190571314423,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 33393
    },
    {
      "epoch": 13.341590091889731,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 33394
    },
    {
      "epoch": 13.341989612465042,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 33395
    },
    {
      "epoch": 13.342389133040351,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 33396
    },
    {
      "epoch": 13.342788653615662,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 33397
    },
    {
      "epoch": 13.34318817419097,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 33398
    },
    {
      "epoch": 13.343587694766281,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 33399
    },
    {
      "epoch": 13.34398721534159,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 33400
    },
    {
      "epoch": 13.344386735916899,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 33401
    },
    {
      "epoch": 13.34478625649221,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 33402
    },
    {
      "epoch": 13.345185777067519,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 33403
    },
    {
      "epoch": 13.34558529764283,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 33404
    },
    {
      "epoch": 13.345984818218138,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 33405
    },
    {
      "epoch": 13.346384338793447,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 33406
    },
    {
      "epoch": 13.346783859368758,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 33407
    },
    {
      "epoch": 13.347183379944067,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 33408
    },
    {
      "epoch": 13.347582900519377,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 33409
    },
    {
      "epoch": 13.347982421094686,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 33410
    },
    {
      "epoch": 13.348381941669995,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 33411
    },
    {
      "epoch": 13.348781462245306,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 33412
    },
    {
      "epoch": 13.349180982820615,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 33413
    },
    {
      "epoch": 13.349580503395925,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 33414
    },
    {
      "epoch": 13.349980023971234,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 33415
    },
    {
      "epoch": 13.350379544546545,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 33416
    },
    {
      "epoch": 13.350779065121854,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 33417
    },
    {
      "epoch": 13.351178585697163,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 33418
    },
    {
      "epoch": 13.351578106272473,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 33419
    },
    {
      "epoch": 13.351977626847782,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 33420
    },
    {
      "epoch": 13.352377147423093,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 33421
    },
    {
      "epoch": 13.352776667998402,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 33422
    },
    {
      "epoch": 13.35317618857371,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 33423
    },
    {
      "epoch": 13.353575709149021,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 33424
    },
    {
      "epoch": 13.35397522972433,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 33425
    },
    {
      "epoch": 13.354374750299641,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 33426
    },
    {
      "epoch": 13.35477427087495,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 33427
    },
    {
      "epoch": 13.355173791450259,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 33428
    },
    {
      "epoch": 13.35557331202557,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 33429
    },
    {
      "epoch": 13.355972832600878,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 33430
    },
    {
      "epoch": 13.356372353176189,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 33431
    },
    {
      "epoch": 13.356771873751498,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 33432
    },
    {
      "epoch": 13.357171394326809,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 33433
    },
    {
      "epoch": 13.357570914902118,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 33434
    },
    {
      "epoch": 13.357970435477426,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 33435
    },
    {
      "epoch": 13.358369956052737,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 33436
    },
    {
      "epoch": 13.358769476628046,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 33437
    },
    {
      "epoch": 13.359168997203357,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 33438
    },
    {
      "epoch": 13.359568517778666,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 33439
    },
    {
      "epoch": 13.359968038353975,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 33440
    },
    {
      "epoch": 13.360367558929285,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 33441
    },
    {
      "epoch": 13.360767079504594,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 33442
    },
    {
      "epoch": 13.361166600079905,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 33443
    },
    {
      "epoch": 13.361566120655214,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 33444
    },
    {
      "epoch": 13.361965641230523,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 33445
    },
    {
      "epoch": 13.362365161805833,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 33446
    },
    {
      "epoch": 13.362764682381142,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 33447
    },
    {
      "epoch": 13.363164202956453,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 33448
    },
    {
      "epoch": 13.363563723531762,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 33449
    },
    {
      "epoch": 13.363963244107072,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 33450
    },
    {
      "epoch": 13.364362764682381,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 33451
    },
    {
      "epoch": 13.36476228525769,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 33452
    },
    {
      "epoch": 13.365161805833,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 33453
    },
    {
      "epoch": 13.36556132640831,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 33454
    },
    {
      "epoch": 13.36596084698362,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 33455
    },
    {
      "epoch": 13.36636036755893,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 33456
    },
    {
      "epoch": 13.366759888134238,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 33457
    },
    {
      "epoch": 13.367159408709549,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 33458
    },
    {
      "epoch": 13.367558929284858,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 33459
    },
    {
      "epoch": 13.367958449860168,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0516,
      "step": 33460
    },
    {
      "epoch": 13.368357970435477,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 33461
    },
    {
      "epoch": 13.368757491010786,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 33462
    },
    {
      "epoch": 13.369157011586097,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 33463
    },
    {
      "epoch": 13.369556532161406,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 33464
    },
    {
      "epoch": 13.369956052736717,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 33465
    },
    {
      "epoch": 13.370355573312025,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 33466
    },
    {
      "epoch": 13.370755093887336,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 33467
    },
    {
      "epoch": 13.371154614462645,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 33468
    },
    {
      "epoch": 13.371554135037954,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 33469
    },
    {
      "epoch": 13.371953655613265,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 33470
    },
    {
      "epoch": 13.372353176188573,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 33471
    },
    {
      "epoch": 13.372752696763884,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 33472
    },
    {
      "epoch": 13.373152217339193,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 33473
    },
    {
      "epoch": 13.373551737914502,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 33474
    },
    {
      "epoch": 13.373951258489813,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 33475
    },
    {
      "epoch": 13.374350779065121,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 33476
    },
    {
      "epoch": 13.374750299640432,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 33477
    },
    {
      "epoch": 13.375149820215741,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 33478
    },
    {
      "epoch": 13.37554934079105,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 33479
    },
    {
      "epoch": 13.37594886136636,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 33480
    },
    {
      "epoch": 13.37634838194167,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 33481
    },
    {
      "epoch": 13.37674790251698,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 33482
    },
    {
      "epoch": 13.377147423092289,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0114,
      "step": 33483
    },
    {
      "epoch": 13.3775469436676,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 33484
    },
    {
      "epoch": 13.377946464242909,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 33485
    },
    {
      "epoch": 13.378345984818218,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 33486
    },
    {
      "epoch": 13.378745505393528,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 33487
    },
    {
      "epoch": 13.379145025968837,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 33488
    },
    {
      "epoch": 13.379544546544148,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0125,
      "step": 33489
    },
    {
      "epoch": 13.379944067119457,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 33490
    },
    {
      "epoch": 13.380343587694766,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 33491
    },
    {
      "epoch": 13.380743108270076,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 33492
    },
    {
      "epoch": 13.381142628845385,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 33493
    },
    {
      "epoch": 13.381542149420696,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 33494
    },
    {
      "epoch": 13.381941669996005,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 33495
    },
    {
      "epoch": 13.382341190571314,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 33496
    },
    {
      "epoch": 13.382740711146624,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 33497
    },
    {
      "epoch": 13.383140231721933,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 33498
    },
    {
      "epoch": 13.383539752297244,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 33499
    },
    {
      "epoch": 13.383939272872553,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 33500
    },
    {
      "epoch": 13.384338793447863,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 33501
    },
    {
      "epoch": 13.384738314023172,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 33502
    },
    {
      "epoch": 13.385137834598481,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 33503
    },
    {
      "epoch": 13.385537355173792,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 33504
    },
    {
      "epoch": 13.3859368757491,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 33505
    },
    {
      "epoch": 13.386336396324412,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 33506
    },
    {
      "epoch": 13.38673591689972,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 33507
    },
    {
      "epoch": 13.38713543747503,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 33508
    },
    {
      "epoch": 13.38753495805034,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 33509
    },
    {
      "epoch": 13.387934478625649,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 33510
    },
    {
      "epoch": 13.38833399920096,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 33511
    },
    {
      "epoch": 13.388733519776268,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 33512
    },
    {
      "epoch": 13.389133040351577,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 33513
    },
    {
      "epoch": 13.389532560926888,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 33514
    },
    {
      "epoch": 13.389932081502197,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 33515
    },
    {
      "epoch": 13.390331602077508,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 33516
    },
    {
      "epoch": 13.390731122652817,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 33517
    },
    {
      "epoch": 13.391130643228125,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 33518
    },
    {
      "epoch": 13.391530163803436,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 33519
    },
    {
      "epoch": 13.391929684378745,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 33520
    },
    {
      "epoch": 13.392329204954056,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 33521
    },
    {
      "epoch": 13.392728725529365,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 33522
    },
    {
      "epoch": 13.393128246104675,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 33523
    },
    {
      "epoch": 13.393527766679984,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 33524
    },
    {
      "epoch": 13.393927287255293,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 33525
    },
    {
      "epoch": 13.394326807830604,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 33526
    },
    {
      "epoch": 13.394726328405913,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 33527
    },
    {
      "epoch": 13.395125848981223,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 33528
    },
    {
      "epoch": 13.395525369556532,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 33529
    },
    {
      "epoch": 13.395924890131841,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 33530
    },
    {
      "epoch": 13.396324410707152,
      "grad_norm": 0.036376953125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 33531
    },
    {
      "epoch": 13.39672393128246,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 33532
    },
    {
      "epoch": 13.397123451857771,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 33533
    },
    {
      "epoch": 13.39752297243308,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 33534
    },
    {
      "epoch": 13.397922493008389,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 33535
    },
    {
      "epoch": 13.3983220135837,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 33536
    },
    {
      "epoch": 13.398721534159009,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 33537
    },
    {
      "epoch": 13.39912105473432,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 33538
    },
    {
      "epoch": 13.399520575309628,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 33539
    },
    {
      "epoch": 13.399920095884939,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 33540
    },
    {
      "epoch": 13.400319616460248,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 33541
    },
    {
      "epoch": 13.400719137035557,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 33542
    },
    {
      "epoch": 13.401118657610867,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 33543
    },
    {
      "epoch": 13.401518178186176,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 33544
    },
    {
      "epoch": 13.401917698761487,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 33545
    },
    {
      "epoch": 13.402317219336796,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 33546
    },
    {
      "epoch": 13.402716739912105,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 33547
    },
    {
      "epoch": 13.403116260487415,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 33548
    },
    {
      "epoch": 13.403515781062724,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 33549
    },
    {
      "epoch": 13.403915301638035,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 33550
    },
    {
      "epoch": 13.404314822213344,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 33551
    },
    {
      "epoch": 13.404714342788653,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 33552
    },
    {
      "epoch": 13.405113863363963,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 33553
    },
    {
      "epoch": 13.405513383939272,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 33554
    },
    {
      "epoch": 13.405912904514583,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 33555
    },
    {
      "epoch": 13.406312425089892,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 33556
    },
    {
      "epoch": 13.4067119456652,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 33557
    },
    {
      "epoch": 13.407111466240512,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 33558
    },
    {
      "epoch": 13.40751098681582,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 33559
    },
    {
      "epoch": 13.407910507391131,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 33560
    },
    {
      "epoch": 13.40831002796644,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 33561
    },
    {
      "epoch": 13.40870954854175,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 33562
    },
    {
      "epoch": 13.40910906911706,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 33563
    },
    {
      "epoch": 13.409508589692368,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 33564
    },
    {
      "epoch": 13.40990811026768,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 33565
    },
    {
      "epoch": 13.410307630842988,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 33566
    },
    {
      "epoch": 13.410707151418299,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 33567
    },
    {
      "epoch": 13.411106671993608,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 33568
    },
    {
      "epoch": 13.411506192568917,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 33569
    },
    {
      "epoch": 13.411905713144227,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 33570
    },
    {
      "epoch": 13.412305233719536,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 33571
    },
    {
      "epoch": 13.412704754294847,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 33572
    },
    {
      "epoch": 13.413104274870156,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 33573
    },
    {
      "epoch": 13.413503795445465,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 33574
    },
    {
      "epoch": 13.413903316020775,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 33575
    },
    {
      "epoch": 13.414302836596084,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 33576
    },
    {
      "epoch": 13.414702357171395,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 33577
    },
    {
      "epoch": 13.415101877746704,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 33578
    },
    {
      "epoch": 13.415501398322014,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 33579
    },
    {
      "epoch": 13.415900918897323,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 33580
    },
    {
      "epoch": 13.416300439472632,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 33581
    },
    {
      "epoch": 13.416699960047943,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 33582
    },
    {
      "epoch": 13.417099480623252,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 33583
    },
    {
      "epoch": 13.417499001198562,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 33584
    },
    {
      "epoch": 13.417898521773871,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 33585
    },
    {
      "epoch": 13.41829804234918,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 33586
    },
    {
      "epoch": 13.418697562924491,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 33587
    },
    {
      "epoch": 13.4190970834998,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 33588
    },
    {
      "epoch": 13.41949660407511,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 33589
    },
    {
      "epoch": 13.41989612465042,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 33590
    },
    {
      "epoch": 13.420295645225728,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 33591
    },
    {
      "epoch": 13.420695165801039,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 33592
    },
    {
      "epoch": 13.421094686376348,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 33593
    },
    {
      "epoch": 13.421494206951659,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 33594
    },
    {
      "epoch": 13.421893727526967,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 33595
    },
    {
      "epoch": 13.422293248102278,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 33596
    },
    {
      "epoch": 13.422692768677587,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 33597
    },
    {
      "epoch": 13.423092289252896,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 33598
    },
    {
      "epoch": 13.423491809828207,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 33599
    },
    {
      "epoch": 13.423891330403515,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 33600
    },
    {
      "epoch": 13.424290850978826,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 33601
    },
    {
      "epoch": 13.424690371554135,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 33602
    },
    {
      "epoch": 13.425089892129444,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 33603
    },
    {
      "epoch": 13.425489412704755,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 33604
    },
    {
      "epoch": 13.425888933280063,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 33605
    },
    {
      "epoch": 13.426288453855374,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 33606
    },
    {
      "epoch": 13.426687974430683,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 33607
    },
    {
      "epoch": 13.427087495005992,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 33608
    },
    {
      "epoch": 13.427487015581303,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 33609
    },
    {
      "epoch": 13.427886536156612,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 33610
    },
    {
      "epoch": 13.428286056731922,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 33611
    },
    {
      "epoch": 13.428685577307231,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 33612
    },
    {
      "epoch": 13.429085097882542,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 33613
    },
    {
      "epoch": 13.42948461845785,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 33614
    },
    {
      "epoch": 13.42988413903316,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 33615
    },
    {
      "epoch": 13.43028365960847,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 33616
    },
    {
      "epoch": 13.43068318018378,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 33617
    },
    {
      "epoch": 13.43108270075909,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 33618
    },
    {
      "epoch": 13.431482221334399,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 33619
    },
    {
      "epoch": 13.431881741909708,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 33620
    },
    {
      "epoch": 13.432281262485018,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 33621
    },
    {
      "epoch": 13.432680783060327,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 33622
    },
    {
      "epoch": 13.433080303635638,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 33623
    },
    {
      "epoch": 13.433479824210947,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 33624
    },
    {
      "epoch": 13.433879344786256,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 33625
    },
    {
      "epoch": 13.434278865361566,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 33626
    },
    {
      "epoch": 13.434678385936875,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 33627
    },
    {
      "epoch": 13.435077906512186,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 33628
    },
    {
      "epoch": 13.435477427087495,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 33629
    },
    {
      "epoch": 13.435876947662805,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 33630
    },
    {
      "epoch": 13.436276468238114,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 33631
    },
    {
      "epoch": 13.436675988813423,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 33632
    },
    {
      "epoch": 13.437075509388734,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 33633
    },
    {
      "epoch": 13.437475029964043,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 33634
    },
    {
      "epoch": 13.437874550539354,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 33635
    },
    {
      "epoch": 13.438274071114662,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 33636
    },
    {
      "epoch": 13.438673591689971,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 33637
    },
    {
      "epoch": 13.439073112265282,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 33638
    },
    {
      "epoch": 13.439472632840591,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 33639
    },
    {
      "epoch": 13.439872153415902,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 33640
    },
    {
      "epoch": 13.44027167399121,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 33641
    },
    {
      "epoch": 13.44067119456652,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 33642
    },
    {
      "epoch": 13.44107071514183,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 33643
    },
    {
      "epoch": 13.441470235717139,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 33644
    },
    {
      "epoch": 13.44186975629245,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 33645
    },
    {
      "epoch": 13.442269276867759,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 33646
    },
    {
      "epoch": 13.44266879744307,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 33647
    },
    {
      "epoch": 13.443068318018378,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 33648
    },
    {
      "epoch": 13.443467838593687,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 33649
    },
    {
      "epoch": 13.443867359168998,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 33650
    },
    {
      "epoch": 13.444266879744307,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 33651
    },
    {
      "epoch": 13.444666400319617,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 33652
    },
    {
      "epoch": 13.445065920894926,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 33653
    },
    {
      "epoch": 13.445465441470235,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 33654
    },
    {
      "epoch": 13.445864962045546,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 33655
    },
    {
      "epoch": 13.446264482620855,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 33656
    },
    {
      "epoch": 13.446664003196165,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 33657
    },
    {
      "epoch": 13.447063523771474,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 33658
    },
    {
      "epoch": 13.447463044346783,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 33659
    },
    {
      "epoch": 13.447862564922094,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 33660
    },
    {
      "epoch": 13.448262085497403,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 33661
    },
    {
      "epoch": 13.448661606072713,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 33662
    },
    {
      "epoch": 13.449061126648022,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 33663
    },
    {
      "epoch": 13.449460647223333,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 33664
    },
    {
      "epoch": 13.449860167798642,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 33665
    },
    {
      "epoch": 13.45025968837395,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 33666
    },
    {
      "epoch": 13.450659208949261,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 33667
    },
    {
      "epoch": 13.45105872952457,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 33668
    },
    {
      "epoch": 13.451458250099881,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 33669
    },
    {
      "epoch": 13.45185777067519,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 33670
    },
    {
      "epoch": 13.452257291250499,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 33671
    },
    {
      "epoch": 13.45265681182581,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 33672
    },
    {
      "epoch": 13.453056332401118,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 33673
    },
    {
      "epoch": 13.453455852976429,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 33674
    },
    {
      "epoch": 13.453855373551738,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 33675
    },
    {
      "epoch": 13.454254894127047,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0484,
      "step": 33676
    },
    {
      "epoch": 13.454654414702357,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 33677
    },
    {
      "epoch": 13.455053935277666,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 33678
    },
    {
      "epoch": 13.455453455852977,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 33679
    },
    {
      "epoch": 13.455852976428286,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 33680
    },
    {
      "epoch": 13.456252497003595,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 33681
    },
    {
      "epoch": 13.456652017578906,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 33682
    },
    {
      "epoch": 13.457051538154214,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 33683
    },
    {
      "epoch": 13.457451058729525,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 33684
    },
    {
      "epoch": 13.457850579304834,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 33685
    },
    {
      "epoch": 13.458250099880145,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 33686
    },
    {
      "epoch": 13.458649620455454,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 33687
    },
    {
      "epoch": 13.459049141030762,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 33688
    },
    {
      "epoch": 13.459448661606073,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 33689
    },
    {
      "epoch": 13.459848182181382,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 33690
    },
    {
      "epoch": 13.460247702756693,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 33691
    },
    {
      "epoch": 13.460647223332002,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 33692
    },
    {
      "epoch": 13.46104674390731,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 33693
    },
    {
      "epoch": 13.461446264482621,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 33694
    },
    {
      "epoch": 13.46184578505793,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 33695
    },
    {
      "epoch": 13.46224530563324,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 33696
    },
    {
      "epoch": 13.46264482620855,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 33697
    },
    {
      "epoch": 13.463044346783859,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 33698
    },
    {
      "epoch": 13.46344386735917,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 33699
    },
    {
      "epoch": 13.463843387934478,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 33700
    },
    {
      "epoch": 13.464242908509789,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 33701
    },
    {
      "epoch": 13.464642429085098,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 33702
    },
    {
      "epoch": 13.465041949660408,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 33703
    },
    {
      "epoch": 13.465441470235717,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 33704
    },
    {
      "epoch": 13.465840990811026,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 33705
    },
    {
      "epoch": 13.466240511386337,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 33706
    },
    {
      "epoch": 13.466640031961646,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.047,
      "step": 33707
    },
    {
      "epoch": 13.467039552536956,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 33708
    },
    {
      "epoch": 13.467439073112265,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 33709
    },
    {
      "epoch": 13.467838593687574,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 33710
    },
    {
      "epoch": 13.468238114262885,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 33711
    },
    {
      "epoch": 13.468637634838194,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 33712
    },
    {
      "epoch": 13.469037155413504,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 33713
    },
    {
      "epoch": 13.469436675988813,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 33714
    },
    {
      "epoch": 13.469836196564122,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 33715
    },
    {
      "epoch": 13.470235717139433,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 33716
    },
    {
      "epoch": 13.470635237714742,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 33717
    },
    {
      "epoch": 13.471034758290052,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 33718
    },
    {
      "epoch": 13.471434278865361,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 33719
    },
    {
      "epoch": 13.471833799440672,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 33720
    },
    {
      "epoch": 13.472233320015981,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 33721
    },
    {
      "epoch": 13.47263284059129,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 33722
    },
    {
      "epoch": 13.4730323611666,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 33723
    },
    {
      "epoch": 13.47343188174191,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 33724
    },
    {
      "epoch": 13.47383140231722,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 33725
    },
    {
      "epoch": 13.474230922892529,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 33726
    },
    {
      "epoch": 13.474630443467838,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 33727
    },
    {
      "epoch": 13.475029964043149,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 33728
    },
    {
      "epoch": 13.475429484618457,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 33729
    },
    {
      "epoch": 13.475829005193768,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 33730
    },
    {
      "epoch": 13.476228525769077,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 33731
    },
    {
      "epoch": 13.476628046344386,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 33732
    },
    {
      "epoch": 13.477027566919697,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 33733
    },
    {
      "epoch": 13.477427087495006,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 33734
    },
    {
      "epoch": 13.477826608070316,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 33735
    },
    {
      "epoch": 13.478226128645625,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 33736
    },
    {
      "epoch": 13.478625649220934,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 33737
    },
    {
      "epoch": 13.479025169796245,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 33738
    },
    {
      "epoch": 13.479424690371554,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 33739
    },
    {
      "epoch": 13.479824210946864,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 33740
    },
    {
      "epoch": 13.480223731522173,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 33741
    },
    {
      "epoch": 13.480623252097484,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 33742
    },
    {
      "epoch": 13.481022772672793,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 33743
    },
    {
      "epoch": 13.481422293248102,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 33744
    },
    {
      "epoch": 13.481821813823412,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 33745
    },
    {
      "epoch": 13.482221334398721,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 33746
    },
    {
      "epoch": 13.482620854974032,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 33747
    },
    {
      "epoch": 13.48302037554934,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 33748
    },
    {
      "epoch": 13.48341989612465,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 33749
    },
    {
      "epoch": 13.48381941669996,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 33750
    },
    {
      "epoch": 13.48421893727527,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 33751
    },
    {
      "epoch": 13.48461845785058,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 33752
    },
    {
      "epoch": 13.485017978425889,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 33753
    },
    {
      "epoch": 13.485417499001198,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 33754
    },
    {
      "epoch": 13.485817019576508,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 33755
    },
    {
      "epoch": 13.486216540151817,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 33756
    },
    {
      "epoch": 13.486616060727128,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 33757
    },
    {
      "epoch": 13.487015581302437,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 33758
    },
    {
      "epoch": 13.487415101877748,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 33759
    },
    {
      "epoch": 13.487814622453056,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 33760
    },
    {
      "epoch": 13.488214143028365,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 33761
    },
    {
      "epoch": 13.488613663603676,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 33762
    },
    {
      "epoch": 13.489013184178985,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 33763
    },
    {
      "epoch": 13.489412704754296,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 33764
    },
    {
      "epoch": 13.489812225329604,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 33765
    },
    {
      "epoch": 13.490211745904913,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 33766
    },
    {
      "epoch": 13.490611266480224,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 33767
    },
    {
      "epoch": 13.491010787055533,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 33768
    },
    {
      "epoch": 13.491410307630844,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 33769
    },
    {
      "epoch": 13.491809828206152,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 33770
    },
    {
      "epoch": 13.492209348781461,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 33771
    },
    {
      "epoch": 13.492608869356772,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 33772
    },
    {
      "epoch": 13.493008389932081,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 33773
    },
    {
      "epoch": 13.493407910507392,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 33774
    },
    {
      "epoch": 13.4938074310827,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0121,
      "step": 33775
    },
    {
      "epoch": 13.494206951658011,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 33776
    },
    {
      "epoch": 13.49460647223332,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 33777
    },
    {
      "epoch": 13.495005992808629,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 33778
    },
    {
      "epoch": 13.49540551338394,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 33779
    },
    {
      "epoch": 13.495805033959249,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 33780
    },
    {
      "epoch": 13.49620455453456,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 33781
    },
    {
      "epoch": 13.496604075109868,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 33782
    },
    {
      "epoch": 13.497003595685177,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 33783
    },
    {
      "epoch": 13.497403116260488,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 33784
    },
    {
      "epoch": 13.497802636835797,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 33785
    },
    {
      "epoch": 13.498202157411107,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 33786
    },
    {
      "epoch": 13.498601677986416,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 33787
    },
    {
      "epoch": 13.499001198561725,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 33788
    },
    {
      "epoch": 13.499400719137036,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 33789
    },
    {
      "epoch": 13.499800239712345,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 33790
    },
    {
      "epoch": 13.500199760287655,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 33791
    },
    {
      "epoch": 13.500599280862964,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 33792
    },
    {
      "epoch": 13.500998801438275,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 33793
    },
    {
      "epoch": 13.501398322013584,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 33794
    },
    {
      "epoch": 13.501797842588893,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 33795
    },
    {
      "epoch": 13.502197363164203,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 33796
    },
    {
      "epoch": 13.502596883739512,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 33797
    },
    {
      "epoch": 13.502996404314823,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 33798
    },
    {
      "epoch": 13.503395924890132,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 33799
    },
    {
      "epoch": 13.50379544546544,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 33800
    },
    {
      "epoch": 13.504194966040751,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 33801
    },
    {
      "epoch": 13.50459448661606,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 33802
    },
    {
      "epoch": 13.504994007191371,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 33803
    },
    {
      "epoch": 13.50539352776668,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 33804
    },
    {
      "epoch": 13.505793048341989,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 33805
    },
    {
      "epoch": 13.5061925689173,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 33806
    },
    {
      "epoch": 13.506592089492608,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 33807
    },
    {
      "epoch": 13.506991610067919,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 33808
    },
    {
      "epoch": 13.507391130643228,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 33809
    },
    {
      "epoch": 13.507790651218539,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 33810
    },
    {
      "epoch": 13.508190171793848,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 33811
    },
    {
      "epoch": 13.508589692369156,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 33812
    },
    {
      "epoch": 13.508989212944467,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 33813
    },
    {
      "epoch": 13.509388733519776,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 33814
    },
    {
      "epoch": 13.509788254095087,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 33815
    },
    {
      "epoch": 13.510187774670396,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 33816
    },
    {
      "epoch": 13.510587295245704,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 33817
    },
    {
      "epoch": 13.510986815821015,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 33818
    },
    {
      "epoch": 13.511386336396324,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 33819
    },
    {
      "epoch": 13.511785856971635,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 33820
    },
    {
      "epoch": 13.512185377546944,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 33821
    },
    {
      "epoch": 13.512584898122252,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 33822
    },
    {
      "epoch": 13.512984418697563,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 33823
    },
    {
      "epoch": 13.513383939272872,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 33824
    },
    {
      "epoch": 13.513783459848183,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 33825
    },
    {
      "epoch": 13.514182980423492,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 33826
    },
    {
      "epoch": 13.514582500998802,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 33827
    },
    {
      "epoch": 13.514982021574111,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 33828
    },
    {
      "epoch": 13.51538154214942,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 33829
    },
    {
      "epoch": 13.51578106272473,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 33830
    },
    {
      "epoch": 13.51618058330004,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 33831
    },
    {
      "epoch": 13.51658010387535,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 33832
    },
    {
      "epoch": 13.51697962445066,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 33833
    },
    {
      "epoch": 13.517379145025968,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 33834
    },
    {
      "epoch": 13.517778665601279,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 33835
    },
    {
      "epoch": 13.518178186176588,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 33836
    },
    {
      "epoch": 13.518577706751898,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 33837
    },
    {
      "epoch": 13.518977227327207,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 33838
    },
    {
      "epoch": 13.519376747902516,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 33839
    },
    {
      "epoch": 13.519776268477827,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 33840
    },
    {
      "epoch": 13.520175789053136,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.053,
      "step": 33841
    },
    {
      "epoch": 13.520575309628446,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 33842
    },
    {
      "epoch": 13.520974830203755,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 33843
    },
    {
      "epoch": 13.521374350779066,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 33844
    },
    {
      "epoch": 13.521773871354375,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 33845
    },
    {
      "epoch": 13.522173391929684,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 33846
    },
    {
      "epoch": 13.522572912504994,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 33847
    },
    {
      "epoch": 13.522972433080303,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 33848
    },
    {
      "epoch": 13.523371953655614,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 33849
    },
    {
      "epoch": 13.523771474230923,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 33850
    },
    {
      "epoch": 13.524170994806232,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 33851
    },
    {
      "epoch": 13.524570515381543,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 33852
    },
    {
      "epoch": 13.524970035956851,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 33853
    },
    {
      "epoch": 13.525369556532162,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 33854
    },
    {
      "epoch": 13.525769077107471,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 33855
    },
    {
      "epoch": 13.52616859768278,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 33856
    },
    {
      "epoch": 13.52656811825809,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 33857
    },
    {
      "epoch": 13.5269676388334,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 33858
    },
    {
      "epoch": 13.52736715940871,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 33859
    },
    {
      "epoch": 13.527766679984019,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 33860
    },
    {
      "epoch": 13.52816620055933,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 33861
    },
    {
      "epoch": 13.528565721134639,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 33862
    },
    {
      "epoch": 13.528965241709948,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 33863
    },
    {
      "epoch": 13.529364762285258,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 33864
    },
    {
      "epoch": 13.529764282860567,
      "grad_norm": 0.035888671875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 33865
    },
    {
      "epoch": 13.530163803435878,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 33866
    },
    {
      "epoch": 13.530563324011187,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 33867
    },
    {
      "epoch": 13.530962844586496,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 33868
    },
    {
      "epoch": 13.531362365161806,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 33869
    },
    {
      "epoch": 13.531761885737115,
      "grad_norm": 0.03662109375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 33870
    },
    {
      "epoch": 13.532161406312426,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 33871
    },
    {
      "epoch": 13.532560926887735,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 33872
    },
    {
      "epoch": 13.532960447463044,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 33873
    },
    {
      "epoch": 13.533359968038354,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 33874
    },
    {
      "epoch": 13.533759488613663,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 33875
    },
    {
      "epoch": 13.534159009188974,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 33876
    },
    {
      "epoch": 13.534558529764283,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 33877
    },
    {
      "epoch": 13.534958050339593,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 33878
    },
    {
      "epoch": 13.535357570914902,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 33879
    },
    {
      "epoch": 13.535757091490211,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 33880
    },
    {
      "epoch": 13.536156612065522,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 33881
    },
    {
      "epoch": 13.53655613264083,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 33882
    },
    {
      "epoch": 13.53695565321614,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 33883
    },
    {
      "epoch": 13.53735517379145,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 33884
    },
    {
      "epoch": 13.53775469436676,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 33885
    },
    {
      "epoch": 13.53815421494207,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 33886
    },
    {
      "epoch": 13.538553735517379,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 33887
    },
    {
      "epoch": 13.53895325609269,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 33888
    },
    {
      "epoch": 13.539352776667998,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 33889
    },
    {
      "epoch": 13.539752297243307,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 33890
    },
    {
      "epoch": 13.540151817818618,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 33891
    },
    {
      "epoch": 13.540551338393927,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 33892
    },
    {
      "epoch": 13.540950858969238,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 33893
    },
    {
      "epoch": 13.541350379544546,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 33894
    },
    {
      "epoch": 13.541749900119855,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 33895
    },
    {
      "epoch": 13.542149420695166,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 33896
    },
    {
      "epoch": 13.542548941270475,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 33897
    },
    {
      "epoch": 13.542948461845786,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 33898
    },
    {
      "epoch": 13.543347982421094,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 33899
    },
    {
      "epoch": 13.543747502996403,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 33900
    },
    {
      "epoch": 13.544147023571714,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 33901
    },
    {
      "epoch": 13.544546544147023,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 33902
    },
    {
      "epoch": 13.544946064722334,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 33903
    },
    {
      "epoch": 13.545345585297643,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 33904
    },
    {
      "epoch": 13.545745105872953,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 33905
    },
    {
      "epoch": 13.546144626448262,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 33906
    },
    {
      "epoch": 13.546544147023571,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 33907
    },
    {
      "epoch": 13.546943667598882,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 33908
    },
    {
      "epoch": 13.54734318817419,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 33909
    },
    {
      "epoch": 13.547742708749501,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 33910
    },
    {
      "epoch": 13.54814222932481,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 33911
    },
    {
      "epoch": 13.548541749900119,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 33912
    },
    {
      "epoch": 13.54894127047543,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 33913
    },
    {
      "epoch": 13.549340791050739,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 33914
    },
    {
      "epoch": 13.54974031162605,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 33915
    },
    {
      "epoch": 13.550139832201358,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 33916
    },
    {
      "epoch": 13.550539352776667,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 33917
    },
    {
      "epoch": 13.550938873351978,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 33918
    },
    {
      "epoch": 13.551338393927287,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 33919
    },
    {
      "epoch": 13.551737914502597,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 33920
    },
    {
      "epoch": 13.552137435077906,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 33921
    },
    {
      "epoch": 13.552536955653217,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 33922
    },
    {
      "epoch": 13.552936476228526,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0056,
      "step": 33923
    },
    {
      "epoch": 13.553335996803835,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 33924
    },
    {
      "epoch": 13.553735517379145,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 33925
    },
    {
      "epoch": 13.554135037954454,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 33926
    },
    {
      "epoch": 13.554534558529765,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 33927
    },
    {
      "epoch": 13.554934079105074,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 33928
    },
    {
      "epoch": 13.555333599680383,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 33929
    },
    {
      "epoch": 13.555733120255693,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 33930
    },
    {
      "epoch": 13.556132640831002,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 33931
    },
    {
      "epoch": 13.556532161406313,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 33932
    },
    {
      "epoch": 13.556931681981622,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 33933
    },
    {
      "epoch": 13.55733120255693,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 33934
    },
    {
      "epoch": 13.557730723132241,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 33935
    },
    {
      "epoch": 13.55813024370755,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 33936
    },
    {
      "epoch": 13.558529764282861,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 33937
    },
    {
      "epoch": 13.55892928485817,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 33938
    },
    {
      "epoch": 13.55932880543348,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 33939
    },
    {
      "epoch": 13.55972832600879,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 33940
    },
    {
      "epoch": 13.560127846584098,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 33941
    },
    {
      "epoch": 13.560527367159409,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0477,
      "step": 33942
    },
    {
      "epoch": 13.560926887734718,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 33943
    },
    {
      "epoch": 13.561326408310029,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 33944
    },
    {
      "epoch": 13.561725928885338,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 33945
    },
    {
      "epoch": 13.562125449460646,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 33946
    },
    {
      "epoch": 13.562524970035957,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 33947
    },
    {
      "epoch": 13.562924490611266,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 33948
    },
    {
      "epoch": 13.563324011186577,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 33949
    },
    {
      "epoch": 13.563723531761886,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 33950
    },
    {
      "epoch": 13.564123052337195,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 33951
    },
    {
      "epoch": 13.564522572912505,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 33952
    },
    {
      "epoch": 13.564922093487814,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 33953
    },
    {
      "epoch": 13.565321614063125,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 33954
    },
    {
      "epoch": 13.565721134638434,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 33955
    },
    {
      "epoch": 13.566120655213744,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 33956
    },
    {
      "epoch": 13.566520175789053,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 33957
    },
    {
      "epoch": 13.566919696364362,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 33958
    },
    {
      "epoch": 13.567319216939673,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 33959
    },
    {
      "epoch": 13.567718737514982,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 33960
    },
    {
      "epoch": 13.568118258090292,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 33961
    },
    {
      "epoch": 13.568517778665601,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 33962
    },
    {
      "epoch": 13.56891729924091,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 33963
    },
    {
      "epoch": 13.56931681981622,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 33964
    },
    {
      "epoch": 13.56971634039153,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 33965
    },
    {
      "epoch": 13.57011586096684,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 33966
    },
    {
      "epoch": 13.57051538154215,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 33967
    },
    {
      "epoch": 13.570914902117458,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 33968
    },
    {
      "epoch": 13.571314422692769,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 33969
    },
    {
      "epoch": 13.571713943268078,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 33970
    },
    {
      "epoch": 13.572113463843388,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 33971
    },
    {
      "epoch": 13.572512984418697,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 33972
    },
    {
      "epoch": 13.572912504994008,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 33973
    },
    {
      "epoch": 13.573312025569317,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 33974
    },
    {
      "epoch": 13.573711546144626,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 33975
    },
    {
      "epoch": 13.574111066719937,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 33976
    },
    {
      "epoch": 13.574510587295245,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 33977
    },
    {
      "epoch": 13.574910107870556,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 33978
    },
    {
      "epoch": 13.575309628445865,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 33979
    },
    {
      "epoch": 13.575709149021174,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 33980
    },
    {
      "epoch": 13.576108669596485,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 33981
    },
    {
      "epoch": 13.576508190171793,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 33982
    },
    {
      "epoch": 13.576907710747104,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 33983
    },
    {
      "epoch": 13.577307231322413,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 33984
    },
    {
      "epoch": 13.577706751897722,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 33985
    },
    {
      "epoch": 13.578106272473033,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 33986
    },
    {
      "epoch": 13.578505793048341,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 33987
    },
    {
      "epoch": 13.578905313623652,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 33988
    },
    {
      "epoch": 13.579304834198961,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 33989
    },
    {
      "epoch": 13.579704354774272,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 33990
    },
    {
      "epoch": 13.58010387534958,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 33991
    },
    {
      "epoch": 13.58050339592489,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 33992
    },
    {
      "epoch": 13.5809029165002,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 33993
    },
    {
      "epoch": 13.581302437075509,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 33994
    },
    {
      "epoch": 13.58170195765082,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 33995
    },
    {
      "epoch": 13.582101478226129,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 33996
    },
    {
      "epoch": 13.582500998801438,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 33997
    },
    {
      "epoch": 13.582900519376748,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 33998
    },
    {
      "epoch": 13.583300039952057,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 33999
    },
    {
      "epoch": 13.583699560527368,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 34000
    },
    {
      "epoch": 13.584099081102677,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 34001
    },
    {
      "epoch": 13.584498601677986,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 34002
    },
    {
      "epoch": 13.584898122253296,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 34003
    },
    {
      "epoch": 13.585297642828605,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 34004
    },
    {
      "epoch": 13.585697163403916,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 34005
    },
    {
      "epoch": 13.586096683979225,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 34006
    },
    {
      "epoch": 13.586496204554535,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 34007
    },
    {
      "epoch": 13.586895725129844,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 34008
    },
    {
      "epoch": 13.587295245705153,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 34009
    },
    {
      "epoch": 13.587694766280464,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 34010
    },
    {
      "epoch": 13.588094286855773,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 34011
    },
    {
      "epoch": 13.588493807431083,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 34012
    },
    {
      "epoch": 13.588893328006392,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 34013
    },
    {
      "epoch": 13.589292848581701,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 34014
    },
    {
      "epoch": 13.589692369157012,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 34015
    },
    {
      "epoch": 13.59009188973232,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 34016
    },
    {
      "epoch": 13.590491410307632,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 34017
    },
    {
      "epoch": 13.59089093088294,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 34018
    },
    {
      "epoch": 13.59129045145825,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 34019
    },
    {
      "epoch": 13.59168997203356,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 34020
    },
    {
      "epoch": 13.592089492608869,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 34021
    },
    {
      "epoch": 13.59248901318418,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 34022
    },
    {
      "epoch": 13.592888533759488,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 34023
    },
    {
      "epoch": 13.5932880543348,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 34024
    },
    {
      "epoch": 13.593687574910108,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 34025
    },
    {
      "epoch": 13.594087095485417,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 34026
    },
    {
      "epoch": 13.594486616060728,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 34027
    },
    {
      "epoch": 13.594886136636037,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 34028
    },
    {
      "epoch": 13.595285657211347,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 34029
    },
    {
      "epoch": 13.595685177786656,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 34030
    },
    {
      "epoch": 13.596084698361965,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 34031
    },
    {
      "epoch": 13.596484218937276,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 34032
    },
    {
      "epoch": 13.596883739512585,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 34033
    },
    {
      "epoch": 13.597283260087895,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 34034
    },
    {
      "epoch": 13.597682780663204,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 34035
    },
    {
      "epoch": 13.598082301238513,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 34036
    },
    {
      "epoch": 13.598481821813824,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 34037
    },
    {
      "epoch": 13.598881342389133,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 34038
    },
    {
      "epoch": 13.599280862964443,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 34039
    },
    {
      "epoch": 13.599680383539752,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 34040
    },
    {
      "epoch": 13.600079904115063,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 34041
    },
    {
      "epoch": 13.600479424690372,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 34042
    },
    {
      "epoch": 13.60087894526568,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 34043
    },
    {
      "epoch": 13.601278465840991,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 34044
    },
    {
      "epoch": 13.6016779864163,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 34045
    },
    {
      "epoch": 13.602077506991611,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 34046
    },
    {
      "epoch": 13.60247702756692,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 34047
    },
    {
      "epoch": 13.602876548142229,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 34048
    },
    {
      "epoch": 13.60327606871754,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 34049
    },
    {
      "epoch": 13.603675589292848,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 34050
    },
    {
      "epoch": 13.604075109868159,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 34051
    },
    {
      "epoch": 13.604474630443468,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 34052
    },
    {
      "epoch": 13.604874151018777,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 34053
    },
    {
      "epoch": 13.605273671594087,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 34054
    },
    {
      "epoch": 13.605673192169396,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 34055
    },
    {
      "epoch": 13.606072712744707,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 34056
    },
    {
      "epoch": 13.606472233320016,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 34057
    },
    {
      "epoch": 13.606871753895327,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 34058
    },
    {
      "epoch": 13.607271274470635,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 34059
    },
    {
      "epoch": 13.607670795045944,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 34060
    },
    {
      "epoch": 13.608070315621255,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 34061
    },
    {
      "epoch": 13.608469836196564,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 34062
    },
    {
      "epoch": 13.608869356771873,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 34063
    },
    {
      "epoch": 13.609268877347183,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 34064
    },
    {
      "epoch": 13.609668397922492,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 34065
    },
    {
      "epoch": 13.610067918497803,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 34066
    },
    {
      "epoch": 13.610467439073112,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 34067
    },
    {
      "epoch": 13.610866959648423,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 34068
    },
    {
      "epoch": 13.611266480223732,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 34069
    },
    {
      "epoch": 13.61166600079904,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0126,
      "step": 34070
    },
    {
      "epoch": 13.612065521374351,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 34071
    },
    {
      "epoch": 13.61246504194966,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 34072
    },
    {
      "epoch": 13.61286456252497,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 34073
    },
    {
      "epoch": 13.61326408310028,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 34074
    },
    {
      "epoch": 13.613663603675588,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 34075
    },
    {
      "epoch": 13.6140631242509,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 34076
    },
    {
      "epoch": 13.614462644826208,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 34077
    },
    {
      "epoch": 13.614862165401519,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 34078
    },
    {
      "epoch": 13.615261685976828,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 34079
    },
    {
      "epoch": 13.615661206552137,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 34080
    },
    {
      "epoch": 13.616060727127447,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 34081
    },
    {
      "epoch": 13.616460247702756,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 34082
    },
    {
      "epoch": 13.616859768278067,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 34083
    },
    {
      "epoch": 13.617259288853376,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 34084
    },
    {
      "epoch": 13.617658809428686,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 34085
    },
    {
      "epoch": 13.618058330003995,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 34086
    },
    {
      "epoch": 13.618457850579304,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 34087
    },
    {
      "epoch": 13.618857371154615,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 34088
    },
    {
      "epoch": 13.619256891729924,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 34089
    },
    {
      "epoch": 13.619656412305234,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 34090
    },
    {
      "epoch": 13.620055932880543,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 34091
    },
    {
      "epoch": 13.620455453455852,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 34092
    },
    {
      "epoch": 13.620854974031163,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 34093
    },
    {
      "epoch": 13.621254494606472,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 34094
    },
    {
      "epoch": 13.621654015181782,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 34095
    },
    {
      "epoch": 13.622053535757091,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 34096
    },
    {
      "epoch": 13.6224530563324,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 34097
    },
    {
      "epoch": 13.622852576907711,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 34098
    },
    {
      "epoch": 13.62325209748302,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 34099
    },
    {
      "epoch": 13.62365161805833,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 34100
    },
    {
      "epoch": 13.62405113863364,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 34101
    },
    {
      "epoch": 13.62445065920895,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 34102
    },
    {
      "epoch": 13.624850179784259,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 34103
    },
    {
      "epoch": 13.625249700359568,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 34104
    },
    {
      "epoch": 13.625649220934879,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 34105
    },
    {
      "epoch": 13.626048741510187,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 34106
    },
    {
      "epoch": 13.626448262085498,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 34107
    },
    {
      "epoch": 13.626847782660807,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 34108
    },
    {
      "epoch": 13.627247303236116,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 34109
    },
    {
      "epoch": 13.627646823811427,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 34110
    },
    {
      "epoch": 13.628046344386735,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 34111
    },
    {
      "epoch": 13.628445864962046,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 34112
    },
    {
      "epoch": 13.628845385537355,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 34113
    },
    {
      "epoch": 13.629244906112664,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 34114
    },
    {
      "epoch": 13.629644426687975,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 34115
    },
    {
      "epoch": 13.630043947263283,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 34116
    },
    {
      "epoch": 13.630443467838594,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 34117
    },
    {
      "epoch": 13.630842988413903,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 34118
    },
    {
      "epoch": 13.631242508989214,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 34119
    },
    {
      "epoch": 13.631642029564523,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 34120
    },
    {
      "epoch": 13.632041550139832,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 34121
    },
    {
      "epoch": 13.632441070715142,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 34122
    },
    {
      "epoch": 13.632840591290451,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 34123
    },
    {
      "epoch": 13.633240111865762,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 34124
    },
    {
      "epoch": 13.63363963244107,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 34125
    },
    {
      "epoch": 13.63403915301638,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 34126
    },
    {
      "epoch": 13.63443867359169,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 34127
    },
    {
      "epoch": 13.634838194167,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 34128
    },
    {
      "epoch": 13.63523771474231,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 34129
    },
    {
      "epoch": 13.635637235317619,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 34130
    },
    {
      "epoch": 13.636036755892928,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 34131
    },
    {
      "epoch": 13.636436276468238,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 34132
    },
    {
      "epoch": 13.636835797043547,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 34133
    },
    {
      "epoch": 13.637235317618858,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 34134
    },
    {
      "epoch": 13.637634838194167,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 34135
    },
    {
      "epoch": 13.638034358769477,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 34136
    },
    {
      "epoch": 13.638433879344786,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 34137
    },
    {
      "epoch": 13.638833399920095,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 34138
    },
    {
      "epoch": 13.639232920495406,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 34139
    },
    {
      "epoch": 13.639632441070715,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.011,
      "step": 34140
    },
    {
      "epoch": 13.640031961646025,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 34141
    },
    {
      "epoch": 13.640431482221334,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 34142
    },
    {
      "epoch": 13.640831002796643,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 34143
    },
    {
      "epoch": 13.641230523371954,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 34144
    },
    {
      "epoch": 13.641630043947263,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 34145
    },
    {
      "epoch": 13.642029564522574,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 34146
    },
    {
      "epoch": 13.642429085097882,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 34147
    },
    {
      "epoch": 13.642828605673191,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 34148
    },
    {
      "epoch": 13.643228126248502,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 34149
    },
    {
      "epoch": 13.643627646823811,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 34150
    },
    {
      "epoch": 13.644027167399122,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 34151
    },
    {
      "epoch": 13.64442668797443,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 34152
    },
    {
      "epoch": 13.644826208549741,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 34153
    },
    {
      "epoch": 13.64522572912505,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 34154
    },
    {
      "epoch": 13.645625249700359,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 34155
    },
    {
      "epoch": 13.64602477027567,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 34156
    },
    {
      "epoch": 13.646424290850979,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 34157
    },
    {
      "epoch": 13.64682381142629,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 34158
    },
    {
      "epoch": 13.647223332001598,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 34159
    },
    {
      "epoch": 13.647622852576907,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 34160
    },
    {
      "epoch": 13.648022373152218,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 34161
    },
    {
      "epoch": 13.648421893727527,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 34162
    },
    {
      "epoch": 13.648821414302837,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 34163
    },
    {
      "epoch": 13.649220934878146,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 34164
    },
    {
      "epoch": 13.649620455453455,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 34165
    },
    {
      "epoch": 13.650019976028766,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 34166
    },
    {
      "epoch": 13.650419496604075,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 34167
    },
    {
      "epoch": 13.650819017179385,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 34168
    },
    {
      "epoch": 13.651218537754694,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 34169
    },
    {
      "epoch": 13.651618058330005,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 34170
    },
    {
      "epoch": 13.652017578905314,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 34171
    },
    {
      "epoch": 13.652417099480623,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 34172
    },
    {
      "epoch": 13.652816620055933,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 34173
    },
    {
      "epoch": 13.653216140631242,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 34174
    },
    {
      "epoch": 13.653615661206553,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 34175
    },
    {
      "epoch": 13.654015181781862,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 34176
    },
    {
      "epoch": 13.65441470235717,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 34177
    },
    {
      "epoch": 13.654814222932481,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 34178
    },
    {
      "epoch": 13.65521374350779,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 34179
    },
    {
      "epoch": 13.655613264083101,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 34180
    },
    {
      "epoch": 13.65601278465841,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 34181
    },
    {
      "epoch": 13.656412305233719,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 34182
    },
    {
      "epoch": 13.65681182580903,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 34183
    },
    {
      "epoch": 13.657211346384338,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 34184
    },
    {
      "epoch": 13.657610866959649,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 34185
    },
    {
      "epoch": 13.658010387534958,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 34186
    },
    {
      "epoch": 13.658409908110269,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 34187
    },
    {
      "epoch": 13.658809428685577,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 34188
    },
    {
      "epoch": 13.659208949260886,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 34189
    },
    {
      "epoch": 13.659608469836197,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 34190
    },
    {
      "epoch": 13.660007990411506,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 34191
    },
    {
      "epoch": 13.660407510986817,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 34192
    },
    {
      "epoch": 13.660807031562126,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 34193
    },
    {
      "epoch": 13.661206552137434,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 34194
    },
    {
      "epoch": 13.661606072712745,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 34195
    },
    {
      "epoch": 13.662005593288054,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 34196
    },
    {
      "epoch": 13.662405113863365,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 34197
    },
    {
      "epoch": 13.662804634438674,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 34198
    },
    {
      "epoch": 13.663204155013982,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 34199
    },
    {
      "epoch": 13.663603675589293,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 34200
    },
    {
      "epoch": 13.664003196164602,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 34201
    },
    {
      "epoch": 13.664402716739913,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 34202
    },
    {
      "epoch": 13.664802237315222,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 34203
    },
    {
      "epoch": 13.665201757890532,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 34204
    },
    {
      "epoch": 13.665601278465841,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 34205
    },
    {
      "epoch": 13.66600079904115,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 34206
    },
    {
      "epoch": 13.66640031961646,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 34207
    },
    {
      "epoch": 13.66679984019177,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 34208
    },
    {
      "epoch": 13.66719936076708,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 34209
    },
    {
      "epoch": 13.66759888134239,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 34210
    },
    {
      "epoch": 13.667998401917698,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 34211
    },
    {
      "epoch": 13.668397922493009,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 34212
    },
    {
      "epoch": 13.668797443068318,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 34213
    },
    {
      "epoch": 13.669196963643628,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 34214
    },
    {
      "epoch": 13.669596484218937,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 34215
    },
    {
      "epoch": 13.669996004794246,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 34216
    },
    {
      "epoch": 13.670395525369557,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 34217
    },
    {
      "epoch": 13.670795045944866,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 34218
    },
    {
      "epoch": 13.671194566520176,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 34219
    },
    {
      "epoch": 13.671594087095485,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 34220
    },
    {
      "epoch": 13.671993607670796,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 34221
    },
    {
      "epoch": 13.672393128246105,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 34222
    },
    {
      "epoch": 13.672792648821414,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 34223
    },
    {
      "epoch": 13.673192169396724,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 34224
    },
    {
      "epoch": 13.673591689972033,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 34225
    },
    {
      "epoch": 13.673991210547344,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 34226
    },
    {
      "epoch": 13.674390731122653,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 34227
    },
    {
      "epoch": 13.674790251697962,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 34228
    },
    {
      "epoch": 13.675189772273272,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 34229
    },
    {
      "epoch": 13.675589292848581,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 34230
    },
    {
      "epoch": 13.675988813423892,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 34231
    },
    {
      "epoch": 13.676388333999201,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 34232
    },
    {
      "epoch": 13.67678785457451,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 34233
    },
    {
      "epoch": 13.67718737514982,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 34234
    },
    {
      "epoch": 13.67758689572513,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 34235
    },
    {
      "epoch": 13.67798641630044,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 34236
    },
    {
      "epoch": 13.678385936875749,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 34237
    },
    {
      "epoch": 13.67878545745106,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 34238
    },
    {
      "epoch": 13.679184978026369,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 34239
    },
    {
      "epoch": 13.679584498601677,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 34240
    },
    {
      "epoch": 13.679984019176988,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 34241
    },
    {
      "epoch": 13.680383539752297,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 34242
    },
    {
      "epoch": 13.680783060327606,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 34243
    },
    {
      "epoch": 13.681182580902917,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 34244
    },
    {
      "epoch": 13.681582101478226,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 34245
    },
    {
      "epoch": 13.681981622053536,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 34246
    },
    {
      "epoch": 13.682381142628845,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 34247
    },
    {
      "epoch": 13.682780663204156,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 34248
    },
    {
      "epoch": 13.683180183779465,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 34249
    },
    {
      "epoch": 13.683579704354774,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 34250
    },
    {
      "epoch": 13.683979224930084,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 34251
    },
    {
      "epoch": 13.684378745505393,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 34252
    },
    {
      "epoch": 13.684778266080704,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 34253
    },
    {
      "epoch": 13.685177786656013,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 34254
    },
    {
      "epoch": 13.685577307231322,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 34255
    },
    {
      "epoch": 13.685976827806632,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 34256
    },
    {
      "epoch": 13.686376348381941,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 34257
    },
    {
      "epoch": 13.686775868957252,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 34258
    },
    {
      "epoch": 13.68717538953256,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 34259
    },
    {
      "epoch": 13.68757491010787,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 34260
    },
    {
      "epoch": 13.68797443068318,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 34261
    },
    {
      "epoch": 13.68837395125849,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 34262
    },
    {
      "epoch": 13.6887734718338,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 34263
    },
    {
      "epoch": 13.689172992409109,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 34264
    },
    {
      "epoch": 13.68957251298442,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 34265
    },
    {
      "epoch": 13.689972033559728,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 34266
    },
    {
      "epoch": 13.690371554135037,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 34267
    },
    {
      "epoch": 13.690771074710348,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 34268
    },
    {
      "epoch": 13.691170595285657,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 34269
    },
    {
      "epoch": 13.691570115860968,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 34270
    },
    {
      "epoch": 13.691969636436276,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 34271
    },
    {
      "epoch": 13.692369157011585,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 34272
    },
    {
      "epoch": 13.692768677586896,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 34273
    },
    {
      "epoch": 13.693168198162205,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 34274
    },
    {
      "epoch": 13.693567718737516,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 34275
    },
    {
      "epoch": 13.693967239312824,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 34276
    },
    {
      "epoch": 13.694366759888133,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 34277
    },
    {
      "epoch": 13.694766280463444,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 34278
    },
    {
      "epoch": 13.695165801038753,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 34279
    },
    {
      "epoch": 13.695565321614064,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 34280
    },
    {
      "epoch": 13.695964842189372,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 34281
    },
    {
      "epoch": 13.696364362764683,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0058,
      "step": 34282
    },
    {
      "epoch": 13.696763883339992,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 34283
    },
    {
      "epoch": 13.697163403915301,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 34284
    },
    {
      "epoch": 13.697562924490612,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 34285
    },
    {
      "epoch": 13.69796244506592,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 34286
    },
    {
      "epoch": 13.698361965641231,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 34287
    },
    {
      "epoch": 13.69876148621654,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 34288
    },
    {
      "epoch": 13.699161006791849,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 34289
    },
    {
      "epoch": 13.69956052736716,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 34290
    },
    {
      "epoch": 13.699960047942469,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 34291
    },
    {
      "epoch": 13.70035956851778,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 34292
    },
    {
      "epoch": 13.700759089093088,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 34293
    },
    {
      "epoch": 13.701158609668397,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 34294
    },
    {
      "epoch": 13.701558130243708,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 34295
    },
    {
      "epoch": 13.701957650819017,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 34296
    },
    {
      "epoch": 13.702357171394327,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 34297
    },
    {
      "epoch": 13.702756691969636,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 34298
    },
    {
      "epoch": 13.703156212544947,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 34299
    },
    {
      "epoch": 13.703555733120256,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 34300
    },
    {
      "epoch": 13.703955253695565,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 34301
    },
    {
      "epoch": 13.704354774270875,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 34302
    },
    {
      "epoch": 13.704754294846184,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 34303
    },
    {
      "epoch": 13.705153815421495,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 34304
    },
    {
      "epoch": 13.705553335996804,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 34305
    },
    {
      "epoch": 13.705952856572113,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 34306
    },
    {
      "epoch": 13.706352377147423,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 34307
    },
    {
      "epoch": 13.706751897722732,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 34308
    },
    {
      "epoch": 13.707151418298043,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 34309
    },
    {
      "epoch": 13.707550938873352,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 34310
    },
    {
      "epoch": 13.70795045944866,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 34311
    },
    {
      "epoch": 13.708349980023971,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 34312
    },
    {
      "epoch": 13.70874950059928,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 34313
    },
    {
      "epoch": 13.709149021174591,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 34314
    },
    {
      "epoch": 13.7095485417499,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 34315
    },
    {
      "epoch": 13.70994806232521,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0128,
      "step": 34316
    },
    {
      "epoch": 13.71034758290052,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 34317
    },
    {
      "epoch": 13.710747103475828,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 34318
    },
    {
      "epoch": 13.711146624051139,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 34319
    },
    {
      "epoch": 13.711546144626448,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 34320
    },
    {
      "epoch": 13.711945665201759,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 34321
    },
    {
      "epoch": 13.712345185777068,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 34322
    },
    {
      "epoch": 13.712744706352376,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 34323
    },
    {
      "epoch": 13.713144226927687,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 34324
    },
    {
      "epoch": 13.713543747502996,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 34325
    },
    {
      "epoch": 13.713943268078307,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 34326
    },
    {
      "epoch": 13.714342788653616,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 34327
    },
    {
      "epoch": 13.714742309228924,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 34328
    },
    {
      "epoch": 13.715141829804235,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 34329
    },
    {
      "epoch": 13.715541350379544,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 34330
    },
    {
      "epoch": 13.715940870954855,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 34331
    },
    {
      "epoch": 13.716340391530164,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 34332
    },
    {
      "epoch": 13.716739912105474,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 34333
    },
    {
      "epoch": 13.717139432680783,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 34334
    },
    {
      "epoch": 13.717538953256092,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 34335
    },
    {
      "epoch": 13.717938473831403,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 34336
    },
    {
      "epoch": 13.718337994406712,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 34337
    },
    {
      "epoch": 13.718737514982022,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 34338
    },
    {
      "epoch": 13.719137035557331,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 34339
    },
    {
      "epoch": 13.71953655613264,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0469,
      "step": 34340
    },
    {
      "epoch": 13.71993607670795,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 34341
    },
    {
      "epoch": 13.72033559728326,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 34342
    },
    {
      "epoch": 13.72073511785857,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.048,
      "step": 34343
    },
    {
      "epoch": 13.72113463843388,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 34344
    },
    {
      "epoch": 13.721534159009188,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 34345
    },
    {
      "epoch": 13.721933679584499,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 34346
    },
    {
      "epoch": 13.722333200159808,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 34347
    },
    {
      "epoch": 13.722732720735118,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 34348
    },
    {
      "epoch": 13.723132241310427,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 34349
    },
    {
      "epoch": 13.723531761885738,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 34350
    },
    {
      "epoch": 13.723931282461047,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 34351
    },
    {
      "epoch": 13.724330803036356,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 34352
    },
    {
      "epoch": 13.724730323611666,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 34353
    },
    {
      "epoch": 13.725129844186975,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 34354
    },
    {
      "epoch": 13.725529364762286,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 34355
    },
    {
      "epoch": 13.725928885337595,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 34356
    },
    {
      "epoch": 13.726328405912904,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 34357
    },
    {
      "epoch": 13.726727926488214,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 34358
    },
    {
      "epoch": 13.727127447063523,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 34359
    },
    {
      "epoch": 13.727526967638834,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 34360
    },
    {
      "epoch": 13.727926488214143,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 34361
    },
    {
      "epoch": 13.728326008789452,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 34362
    },
    {
      "epoch": 13.728725529364763,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 34363
    },
    {
      "epoch": 13.729125049940071,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 34364
    },
    {
      "epoch": 13.729524570515382,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 34365
    },
    {
      "epoch": 13.729924091090691,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 34366
    },
    {
      "epoch": 13.730323611666002,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 34367
    },
    {
      "epoch": 13.73072313224131,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 34368
    },
    {
      "epoch": 13.73112265281662,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 34369
    },
    {
      "epoch": 13.73152217339193,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 34370
    },
    {
      "epoch": 13.731921693967239,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 34371
    },
    {
      "epoch": 13.73232121454255,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 34372
    },
    {
      "epoch": 13.732720735117859,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 34373
    },
    {
      "epoch": 13.733120255693168,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 34374
    },
    {
      "epoch": 13.733519776268478,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 34375
    },
    {
      "epoch": 13.733919296843787,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 34376
    },
    {
      "epoch": 13.734318817419098,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 34377
    },
    {
      "epoch": 13.734718337994407,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 34378
    },
    {
      "epoch": 13.735117858569716,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 34379
    },
    {
      "epoch": 13.735517379145026,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 34380
    },
    {
      "epoch": 13.735916899720335,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 34381
    },
    {
      "epoch": 13.736316420295646,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 34382
    },
    {
      "epoch": 13.736715940870955,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 34383
    },
    {
      "epoch": 13.737115461446265,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 34384
    },
    {
      "epoch": 13.737514982021574,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 34385
    },
    {
      "epoch": 13.737914502596883,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 34386
    },
    {
      "epoch": 13.738314023172194,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 34387
    },
    {
      "epoch": 13.738713543747503,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 34388
    },
    {
      "epoch": 13.739113064322813,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 34389
    },
    {
      "epoch": 13.739512584898122,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 34390
    },
    {
      "epoch": 13.739912105473431,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 34391
    },
    {
      "epoch": 13.740311626048742,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 34392
    },
    {
      "epoch": 13.74071114662405,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 34393
    },
    {
      "epoch": 13.741110667199361,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 34394
    },
    {
      "epoch": 13.74151018777467,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 34395
    },
    {
      "epoch": 13.74190970834998,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 34396
    },
    {
      "epoch": 13.74230922892529,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 34397
    },
    {
      "epoch": 13.742708749500599,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 34398
    },
    {
      "epoch": 13.74310827007591,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 34399
    },
    {
      "epoch": 13.743507790651218,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 34400
    },
    {
      "epoch": 13.743907311226529,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 34401
    },
    {
      "epoch": 13.744306831801838,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 34402
    },
    {
      "epoch": 13.744706352377147,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 34403
    },
    {
      "epoch": 13.745105872952458,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 34404
    },
    {
      "epoch": 13.745505393527766,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 34405
    },
    {
      "epoch": 13.745904914103077,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 34406
    },
    {
      "epoch": 13.746304434678386,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 34407
    },
    {
      "epoch": 13.746703955253695,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 34408
    },
    {
      "epoch": 13.747103475829006,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 34409
    },
    {
      "epoch": 13.747502996404315,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 34410
    },
    {
      "epoch": 13.747902516979625,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 34411
    },
    {
      "epoch": 13.748302037554934,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 34412
    },
    {
      "epoch": 13.748701558130243,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 34413
    },
    {
      "epoch": 13.749101078705554,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 34414
    },
    {
      "epoch": 13.749500599280863,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 34415
    },
    {
      "epoch": 13.749900119856173,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 34416
    },
    {
      "epoch": 13.750299640431482,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 34417
    },
    {
      "epoch": 13.750699161006793,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 34418
    },
    {
      "epoch": 13.751098681582102,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 34419
    },
    {
      "epoch": 13.75149820215741,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 34420
    },
    {
      "epoch": 13.751897722732721,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 34421
    },
    {
      "epoch": 13.75229724330803,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 34422
    },
    {
      "epoch": 13.752696763883339,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 34423
    },
    {
      "epoch": 13.75309628445865,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 34424
    },
    {
      "epoch": 13.753495805033959,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 34425
    },
    {
      "epoch": 13.75389532560927,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 34426
    },
    {
      "epoch": 13.754294846184578,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 34427
    },
    {
      "epoch": 13.754694366759889,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 34428
    },
    {
      "epoch": 13.755093887335198,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 34429
    },
    {
      "epoch": 13.755493407910507,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 34430
    },
    {
      "epoch": 13.755892928485817,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 34431
    },
    {
      "epoch": 13.756292449061126,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 34432
    },
    {
      "epoch": 13.756691969636437,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 34433
    },
    {
      "epoch": 13.757091490211746,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 34434
    },
    {
      "epoch": 13.757491010787055,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 34435
    },
    {
      "epoch": 13.757890531362365,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 34436
    },
    {
      "epoch": 13.758290051937674,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 34437
    },
    {
      "epoch": 13.758689572512985,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 34438
    },
    {
      "epoch": 13.759089093088294,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 34439
    },
    {
      "epoch": 13.759488613663603,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 34440
    },
    {
      "epoch": 13.759888134238913,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 34441
    },
    {
      "epoch": 13.760287654814222,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 34442
    },
    {
      "epoch": 13.760687175389533,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 34443
    },
    {
      "epoch": 13.761086695964842,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 34444
    },
    {
      "epoch": 13.761486216540153,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 34445
    },
    {
      "epoch": 13.761885737115461,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 34446
    },
    {
      "epoch": 13.76228525769077,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 34447
    },
    {
      "epoch": 13.762684778266081,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 34448
    },
    {
      "epoch": 13.76308429884139,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 34449
    },
    {
      "epoch": 13.7634838194167,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 34450
    },
    {
      "epoch": 13.76388333999201,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 34451
    },
    {
      "epoch": 13.764282860567318,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 34452
    },
    {
      "epoch": 13.764682381142629,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 34453
    },
    {
      "epoch": 13.765081901717938,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 34454
    },
    {
      "epoch": 13.765481422293249,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 34455
    },
    {
      "epoch": 13.765880942868558,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 34456
    },
    {
      "epoch": 13.766280463443866,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 34457
    },
    {
      "epoch": 13.766679984019177,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 34458
    },
    {
      "epoch": 13.767079504594486,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 34459
    },
    {
      "epoch": 13.767479025169797,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 34460
    },
    {
      "epoch": 13.767878545745106,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 34461
    },
    {
      "epoch": 13.768278066320416,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 34462
    },
    {
      "epoch": 13.768677586895725,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 34463
    },
    {
      "epoch": 13.769077107471034,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 34464
    },
    {
      "epoch": 13.769476628046345,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 34465
    },
    {
      "epoch": 13.769876148621654,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 34466
    },
    {
      "epoch": 13.770275669196964,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 34467
    },
    {
      "epoch": 13.770675189772273,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 34468
    },
    {
      "epoch": 13.771074710347582,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 34469
    },
    {
      "epoch": 13.771474230922893,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 34470
    },
    {
      "epoch": 13.771873751498202,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 34471
    },
    {
      "epoch": 13.772273272073512,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 34472
    },
    {
      "epoch": 13.772672792648821,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 34473
    },
    {
      "epoch": 13.77307231322413,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 34474
    },
    {
      "epoch": 13.77347183379944,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 34475
    },
    {
      "epoch": 13.77387135437475,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 34476
    },
    {
      "epoch": 13.77427087495006,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 34477
    },
    {
      "epoch": 13.77467039552537,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 34478
    },
    {
      "epoch": 13.77506991610068,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 34479
    },
    {
      "epoch": 13.775469436675989,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0476,
      "step": 34480
    },
    {
      "epoch": 13.775868957251298,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 34481
    },
    {
      "epoch": 13.776268477826608,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 34482
    },
    {
      "epoch": 13.776667998401917,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 34483
    },
    {
      "epoch": 13.777067518977228,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 34484
    },
    {
      "epoch": 13.777467039552537,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 34485
    },
    {
      "epoch": 13.777866560127846,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 34486
    },
    {
      "epoch": 13.778266080703157,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 34487
    },
    {
      "epoch": 13.778665601278465,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 34488
    },
    {
      "epoch": 13.779065121853776,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 34489
    },
    {
      "epoch": 13.779464642429085,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 34490
    },
    {
      "epoch": 13.779864163004394,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 34491
    },
    {
      "epoch": 13.780263683579705,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 34492
    },
    {
      "epoch": 13.780663204155013,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 34493
    },
    {
      "epoch": 13.781062724730324,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 34494
    },
    {
      "epoch": 13.781462245305633,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 34495
    },
    {
      "epoch": 13.781861765880944,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 34496
    },
    {
      "epoch": 13.782261286456253,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 34497
    },
    {
      "epoch": 13.782660807031561,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 34498
    },
    {
      "epoch": 13.783060327606872,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 34499
    },
    {
      "epoch": 13.783459848182181,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 34500
    },
    {
      "epoch": 13.783859368757492,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 34501
    },
    {
      "epoch": 13.7842588893328,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 34502
    },
    {
      "epoch": 13.78465840990811,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 34503
    },
    {
      "epoch": 13.78505793048342,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 34504
    },
    {
      "epoch": 13.785457451058729,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 34505
    },
    {
      "epoch": 13.78585697163404,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 34506
    },
    {
      "epoch": 13.786256492209349,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 34507
    },
    {
      "epoch": 13.786656012784658,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0501,
      "step": 34508
    },
    {
      "epoch": 13.787055533359968,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 34509
    },
    {
      "epoch": 13.787455053935277,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 34510
    },
    {
      "epoch": 13.787854574510588,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 34511
    },
    {
      "epoch": 13.788254095085897,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 34512
    },
    {
      "epoch": 13.788653615661207,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 34513
    },
    {
      "epoch": 13.789053136236516,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 34514
    },
    {
      "epoch": 13.789452656811825,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 34515
    },
    {
      "epoch": 13.789852177387136,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 34516
    },
    {
      "epoch": 13.790251697962445,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 34517
    },
    {
      "epoch": 13.790651218537755,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 34518
    },
    {
      "epoch": 13.791050739113064,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 34519
    },
    {
      "epoch": 13.791450259688373,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 34520
    },
    {
      "epoch": 13.791849780263684,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 34521
    },
    {
      "epoch": 13.792249300838993,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 34522
    },
    {
      "epoch": 13.792648821414303,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 34523
    },
    {
      "epoch": 13.793048341989612,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 34524
    },
    {
      "epoch": 13.793447862564921,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 34525
    },
    {
      "epoch": 13.793847383140232,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 34526
    },
    {
      "epoch": 13.79424690371554,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 34527
    },
    {
      "epoch": 13.794646424290852,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 34528
    },
    {
      "epoch": 13.79504594486616,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 34529
    },
    {
      "epoch": 13.795445465441471,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 34530
    },
    {
      "epoch": 13.79584498601678,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 34531
    },
    {
      "epoch": 13.796244506592089,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 34532
    },
    {
      "epoch": 13.7966440271674,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 34533
    },
    {
      "epoch": 13.797043547742708,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 34534
    },
    {
      "epoch": 13.79744306831802,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 34535
    },
    {
      "epoch": 13.797842588893328,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 34536
    },
    {
      "epoch": 13.798242109468637,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 34537
    },
    {
      "epoch": 13.798641630043948,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 34538
    },
    {
      "epoch": 13.799041150619257,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 34539
    },
    {
      "epoch": 13.799440671194567,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 34540
    },
    {
      "epoch": 13.799840191769876,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 34541
    },
    {
      "epoch": 13.800239712345185,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 34542
    },
    {
      "epoch": 13.800639232920496,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 34543
    },
    {
      "epoch": 13.801038753495805,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 34544
    },
    {
      "epoch": 13.801438274071115,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 34545
    },
    {
      "epoch": 13.801837794646424,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 34546
    },
    {
      "epoch": 13.802237315221735,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 34547
    },
    {
      "epoch": 13.802636835797044,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 34548
    },
    {
      "epoch": 13.803036356372353,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 34549
    },
    {
      "epoch": 13.803435876947663,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 34550
    },
    {
      "epoch": 13.803835397522972,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 34551
    },
    {
      "epoch": 13.804234918098283,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 34552
    },
    {
      "epoch": 13.804634438673592,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 34553
    },
    {
      "epoch": 13.8050339592489,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 34554
    },
    {
      "epoch": 13.805433479824211,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 34555
    },
    {
      "epoch": 13.80583300039952,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 34556
    },
    {
      "epoch": 13.806232520974831,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 34557
    },
    {
      "epoch": 13.80663204155014,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 34558
    },
    {
      "epoch": 13.807031562125449,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 34559
    },
    {
      "epoch": 13.80743108270076,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 34560
    },
    {
      "epoch": 13.807830603276068,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 34561
    },
    {
      "epoch": 13.808230123851379,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 34562
    },
    {
      "epoch": 13.808629644426688,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 34563
    },
    {
      "epoch": 13.809029165001999,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 34564
    },
    {
      "epoch": 13.809428685577307,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 34565
    },
    {
      "epoch": 13.809828206152616,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 34566
    },
    {
      "epoch": 13.810227726727927,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 34567
    },
    {
      "epoch": 13.810627247303236,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 34568
    },
    {
      "epoch": 13.811026767878547,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 34569
    },
    {
      "epoch": 13.811426288453855,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 34570
    },
    {
      "epoch": 13.811825809029164,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 34571
    },
    {
      "epoch": 13.812225329604475,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 34572
    },
    {
      "epoch": 13.812624850179784,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 34573
    },
    {
      "epoch": 13.813024370755095,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 34574
    },
    {
      "epoch": 13.813423891330403,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 34575
    },
    {
      "epoch": 13.813823411905712,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 34576
    },
    {
      "epoch": 13.814222932481023,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 34577
    },
    {
      "epoch": 13.814622453056332,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 34578
    },
    {
      "epoch": 13.815021973631643,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 34579
    },
    {
      "epoch": 13.815421494206952,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 34580
    },
    {
      "epoch": 13.815821014782262,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 34581
    },
    {
      "epoch": 13.816220535357571,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 34582
    },
    {
      "epoch": 13.81662005593288,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 34583
    },
    {
      "epoch": 13.81701957650819,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0478,
      "step": 34584
    },
    {
      "epoch": 13.8174190970835,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 34585
    },
    {
      "epoch": 13.81781861765881,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 34586
    },
    {
      "epoch": 13.81821813823412,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 34587
    },
    {
      "epoch": 13.818617658809428,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 34588
    },
    {
      "epoch": 13.819017179384739,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 34589
    },
    {
      "epoch": 13.819416699960048,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 34590
    },
    {
      "epoch": 13.819816220535358,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 34591
    },
    {
      "epoch": 13.820215741110667,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 34592
    },
    {
      "epoch": 13.820615261685976,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 34593
    },
    {
      "epoch": 13.821014782261287,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 34594
    },
    {
      "epoch": 13.821414302836596,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 34595
    },
    {
      "epoch": 13.821813823411906,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 34596
    },
    {
      "epoch": 13.822213343987215,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 34597
    },
    {
      "epoch": 13.822612864562524,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 34598
    },
    {
      "epoch": 13.823012385137835,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 34599
    },
    {
      "epoch": 13.823411905713144,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 34600
    },
    {
      "epoch": 13.823811426288454,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 34601
    },
    {
      "epoch": 13.824210946863763,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 34602
    },
    {
      "epoch": 13.824610467439072,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 34603
    },
    {
      "epoch": 13.825009988014383,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 34604
    },
    {
      "epoch": 13.825409508589692,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 34605
    },
    {
      "epoch": 13.825809029165002,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 34606
    },
    {
      "epoch": 13.826208549740311,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 34607
    },
    {
      "epoch": 13.826608070315622,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 34608
    },
    {
      "epoch": 13.827007590890931,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 34609
    },
    {
      "epoch": 13.82740711146624,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 34610
    },
    {
      "epoch": 13.82780663204155,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 34611
    },
    {
      "epoch": 13.82820615261686,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 34612
    },
    {
      "epoch": 13.82860567319217,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0461,
      "step": 34613
    },
    {
      "epoch": 13.829005193767479,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 34614
    },
    {
      "epoch": 13.829404714342788,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 34615
    },
    {
      "epoch": 13.829804234918099,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 34616
    },
    {
      "epoch": 13.830203755493407,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 34617
    },
    {
      "epoch": 13.830603276068718,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 34618
    },
    {
      "epoch": 13.831002796644027,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 34619
    },
    {
      "epoch": 13.831402317219336,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 34620
    },
    {
      "epoch": 13.831801837794647,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 34621
    },
    {
      "epoch": 13.832201358369955,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 34622
    },
    {
      "epoch": 13.832600878945266,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 34623
    },
    {
      "epoch": 13.833000399520575,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 34624
    },
    {
      "epoch": 13.833399920095886,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 34625
    },
    {
      "epoch": 13.833799440671195,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 34626
    },
    {
      "epoch": 13.834198961246504,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 34627
    },
    {
      "epoch": 13.834598481821814,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 34628
    },
    {
      "epoch": 13.834998002397123,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 34629
    },
    {
      "epoch": 13.835397522972434,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 34630
    },
    {
      "epoch": 13.835797043547743,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 34631
    },
    {
      "epoch": 13.836196564123052,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 34632
    },
    {
      "epoch": 13.836596084698362,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 34633
    },
    {
      "epoch": 13.836995605273671,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 34634
    },
    {
      "epoch": 13.837395125848982,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 34635
    },
    {
      "epoch": 13.83779464642429,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 34636
    },
    {
      "epoch": 13.8381941669996,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 34637
    },
    {
      "epoch": 13.83859368757491,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 34638
    },
    {
      "epoch": 13.83899320815022,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 34639
    },
    {
      "epoch": 13.83939272872553,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 34640
    },
    {
      "epoch": 13.839792249300839,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 34641
    },
    {
      "epoch": 13.84019176987615,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 34642
    },
    {
      "epoch": 13.840591290451458,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 34643
    },
    {
      "epoch": 13.840990811026767,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 34644
    },
    {
      "epoch": 13.841390331602078,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 34645
    },
    {
      "epoch": 13.841789852177387,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 34646
    },
    {
      "epoch": 13.842189372752697,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 34647
    },
    {
      "epoch": 13.842588893328006,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 34648
    },
    {
      "epoch": 13.842988413903315,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 34649
    },
    {
      "epoch": 13.843387934478626,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 34650
    },
    {
      "epoch": 13.843787455053935,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 34651
    },
    {
      "epoch": 13.844186975629245,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 34652
    },
    {
      "epoch": 13.844586496204554,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 34653
    },
    {
      "epoch": 13.844986016779863,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 34654
    },
    {
      "epoch": 13.845385537355174,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 34655
    },
    {
      "epoch": 13.845785057930483,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 34656
    },
    {
      "epoch": 13.846184578505794,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 34657
    },
    {
      "epoch": 13.846584099081102,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 34658
    },
    {
      "epoch": 13.846983619656413,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 34659
    },
    {
      "epoch": 13.847383140231722,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 34660
    },
    {
      "epoch": 13.847782660807031,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 34661
    },
    {
      "epoch": 13.848182181382342,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 34662
    },
    {
      "epoch": 13.84858170195765,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 34663
    },
    {
      "epoch": 13.848981222532961,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 34664
    },
    {
      "epoch": 13.84938074310827,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 34665
    },
    {
      "epoch": 13.849780263683579,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 34666
    },
    {
      "epoch": 13.85017978425889,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 34667
    },
    {
      "epoch": 13.850579304834199,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 34668
    },
    {
      "epoch": 13.85097882540951,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 34669
    },
    {
      "epoch": 13.851378345984818,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 34670
    },
    {
      "epoch": 13.851777866560127,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 34671
    },
    {
      "epoch": 13.852177387135438,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 34672
    },
    {
      "epoch": 13.852576907710747,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 34673
    },
    {
      "epoch": 13.852976428286057,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 34674
    },
    {
      "epoch": 13.853375948861366,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 34675
    },
    {
      "epoch": 13.853775469436677,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 34676
    },
    {
      "epoch": 13.854174990011986,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 34677
    },
    {
      "epoch": 13.854574510587295,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 34678
    },
    {
      "epoch": 13.854974031162605,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 34679
    },
    {
      "epoch": 13.855373551737914,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 34680
    },
    {
      "epoch": 13.855773072313225,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 34681
    },
    {
      "epoch": 13.856172592888534,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 34682
    },
    {
      "epoch": 13.856572113463843,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 34683
    },
    {
      "epoch": 13.856971634039153,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 34684
    },
    {
      "epoch": 13.857371154614462,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 34685
    },
    {
      "epoch": 13.857770675189773,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 34686
    },
    {
      "epoch": 13.858170195765082,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 34687
    },
    {
      "epoch": 13.85856971634039,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 34688
    },
    {
      "epoch": 13.858969236915701,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 34689
    },
    {
      "epoch": 13.85936875749101,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 34690
    },
    {
      "epoch": 13.859768278066321,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 34691
    },
    {
      "epoch": 13.86016779864163,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 34692
    },
    {
      "epoch": 13.86056731921694,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 34693
    },
    {
      "epoch": 13.86096683979225,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 34694
    },
    {
      "epoch": 13.861366360367558,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 34695
    },
    {
      "epoch": 13.861765880942869,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 34696
    },
    {
      "epoch": 13.862165401518178,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 34697
    },
    {
      "epoch": 13.862564922093489,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 34698
    },
    {
      "epoch": 13.862964442668797,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 34699
    },
    {
      "epoch": 13.863363963244106,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 34700
    },
    {
      "epoch": 13.863763483819417,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 34701
    },
    {
      "epoch": 13.864163004394726,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 34702
    },
    {
      "epoch": 13.864562524970037,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 34703
    },
    {
      "epoch": 13.864962045545346,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 34704
    },
    {
      "epoch": 13.865361566120654,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 34705
    },
    {
      "epoch": 13.865761086695965,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 34706
    },
    {
      "epoch": 13.866160607271274,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 34707
    },
    {
      "epoch": 13.866560127846585,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 34708
    },
    {
      "epoch": 13.866959648421894,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 34709
    },
    {
      "epoch": 13.867359168997204,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 34710
    },
    {
      "epoch": 13.867758689572513,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 34711
    },
    {
      "epoch": 13.868158210147822,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 34712
    },
    {
      "epoch": 13.868557730723133,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 34713
    },
    {
      "epoch": 13.868957251298442,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 34714
    },
    {
      "epoch": 13.869356771873752,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 34715
    },
    {
      "epoch": 13.869756292449061,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 34716
    },
    {
      "epoch": 13.87015581302437,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 34717
    },
    {
      "epoch": 13.87055533359968,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 34718
    },
    {
      "epoch": 13.87095485417499,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 34719
    },
    {
      "epoch": 13.8713543747503,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 34720
    },
    {
      "epoch": 13.87175389532561,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 34721
    },
    {
      "epoch": 13.872153415900918,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 34722
    },
    {
      "epoch": 13.872552936476229,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 34723
    },
    {
      "epoch": 13.872952457051538,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 34724
    },
    {
      "epoch": 13.873351977626848,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 34725
    },
    {
      "epoch": 13.873751498202157,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 34726
    },
    {
      "epoch": 13.874151018777468,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 34727
    },
    {
      "epoch": 13.874550539352777,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 34728
    },
    {
      "epoch": 13.874950059928086,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 34729
    },
    {
      "epoch": 13.875349580503396,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 34730
    },
    {
      "epoch": 13.875749101078705,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 34731
    },
    {
      "epoch": 13.876148621654016,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 34732
    },
    {
      "epoch": 13.876548142229325,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 34733
    },
    {
      "epoch": 13.876947662804634,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 34734
    },
    {
      "epoch": 13.877347183379944,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 34735
    },
    {
      "epoch": 13.877746703955253,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 34736
    },
    {
      "epoch": 13.878146224530564,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 34737
    },
    {
      "epoch": 13.878545745105873,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 34738
    },
    {
      "epoch": 13.878945265681182,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 34739
    },
    {
      "epoch": 13.879344786256492,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 34740
    },
    {
      "epoch": 13.879744306831801,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 34741
    },
    {
      "epoch": 13.880143827407112,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 34742
    },
    {
      "epoch": 13.880543347982421,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 34743
    },
    {
      "epoch": 13.880942868557732,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0125,
      "step": 34744
    },
    {
      "epoch": 13.88134238913304,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 34745
    },
    {
      "epoch": 13.88174190970835,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 34746
    },
    {
      "epoch": 13.88214143028366,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 34747
    },
    {
      "epoch": 13.882540950858969,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 34748
    },
    {
      "epoch": 13.88294047143428,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 34749
    },
    {
      "epoch": 13.883339992009589,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 34750
    },
    {
      "epoch": 13.883739512584897,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 34751
    },
    {
      "epoch": 13.884139033160208,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 34752
    },
    {
      "epoch": 13.884538553735517,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 34753
    },
    {
      "epoch": 13.884938074310828,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 34754
    },
    {
      "epoch": 13.885337594886137,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 34755
    },
    {
      "epoch": 13.885737115461446,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 34756
    },
    {
      "epoch": 13.886136636036756,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 34757
    },
    {
      "epoch": 13.886536156612065,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 34758
    },
    {
      "epoch": 13.886935677187376,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 34759
    },
    {
      "epoch": 13.887335197762685,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 34760
    },
    {
      "epoch": 13.887734718337995,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 34761
    },
    {
      "epoch": 13.888134238913304,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 34762
    },
    {
      "epoch": 13.888533759488613,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 34763
    },
    {
      "epoch": 13.888933280063924,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 34764
    },
    {
      "epoch": 13.889332800639233,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 34765
    },
    {
      "epoch": 13.889732321214543,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 34766
    },
    {
      "epoch": 13.890131841789852,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 34767
    },
    {
      "epoch": 13.890531362365161,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 34768
    },
    {
      "epoch": 13.890930882940472,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 34769
    },
    {
      "epoch": 13.89133040351578,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 34770
    },
    {
      "epoch": 13.891729924091091,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 34771
    },
    {
      "epoch": 13.8921294446664,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 34772
    },
    {
      "epoch": 13.89252896524171,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 34773
    },
    {
      "epoch": 13.89292848581702,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 34774
    },
    {
      "epoch": 13.893328006392329,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 34775
    },
    {
      "epoch": 13.89372752696764,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 34776
    },
    {
      "epoch": 13.894127047542948,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 34777
    },
    {
      "epoch": 13.894526568118257,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 34778
    },
    {
      "epoch": 13.894926088693568,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 34779
    },
    {
      "epoch": 13.895325609268877,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 34780
    },
    {
      "epoch": 13.895725129844188,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 34781
    },
    {
      "epoch": 13.896124650419496,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 34782
    },
    {
      "epoch": 13.896524170994805,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 34783
    },
    {
      "epoch": 13.896923691570116,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 34784
    },
    {
      "epoch": 13.897323212145425,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 34785
    },
    {
      "epoch": 13.897722732720736,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 34786
    },
    {
      "epoch": 13.898122253296044,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 34787
    },
    {
      "epoch": 13.898521773871355,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 34788
    },
    {
      "epoch": 13.898921294446664,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 34789
    },
    {
      "epoch": 13.899320815021973,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 34790
    },
    {
      "epoch": 13.899720335597284,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 34791
    },
    {
      "epoch": 13.900119856172592,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 34792
    },
    {
      "epoch": 13.900519376747903,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 34793
    },
    {
      "epoch": 13.900918897323212,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 34794
    },
    {
      "epoch": 13.901318417898521,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 34795
    },
    {
      "epoch": 13.901717938473832,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 34796
    },
    {
      "epoch": 13.90211745904914,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0131,
      "step": 34797
    },
    {
      "epoch": 13.902516979624451,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 34798
    },
    {
      "epoch": 13.90291650019976,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 34799
    },
    {
      "epoch": 13.903316020775069,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0121,
      "step": 34800
    },
    {
      "epoch": 13.90371554135038,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 34801
    },
    {
      "epoch": 13.904115061925689,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 34802
    },
    {
      "epoch": 13.904514582501,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 34803
    },
    {
      "epoch": 13.904914103076308,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 34804
    },
    {
      "epoch": 13.905313623651619,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 34805
    },
    {
      "epoch": 13.905713144226928,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 34806
    },
    {
      "epoch": 13.906112664802237,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 34807
    },
    {
      "epoch": 13.906512185377547,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 34808
    },
    {
      "epoch": 13.906911705952856,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 34809
    },
    {
      "epoch": 13.907311226528167,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 34810
    },
    {
      "epoch": 13.907710747103476,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 34811
    },
    {
      "epoch": 13.908110267678785,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 34812
    },
    {
      "epoch": 13.908509788254095,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 34813
    },
    {
      "epoch": 13.908909308829404,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 34814
    },
    {
      "epoch": 13.909308829404715,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 34815
    },
    {
      "epoch": 13.909708349980024,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 34816
    },
    {
      "epoch": 13.910107870555333,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 34817
    },
    {
      "epoch": 13.910507391130643,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 34818
    },
    {
      "epoch": 13.910906911705952,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 34819
    },
    {
      "epoch": 13.911306432281263,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 34820
    },
    {
      "epoch": 13.911705952856572,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 34821
    },
    {
      "epoch": 13.912105473431883,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 34822
    },
    {
      "epoch": 13.912504994007191,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 34823
    },
    {
      "epoch": 13.9129045145825,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 34824
    },
    {
      "epoch": 13.913304035157811,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 34825
    },
    {
      "epoch": 13.91370355573312,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 34826
    },
    {
      "epoch": 13.91410307630843,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 34827
    },
    {
      "epoch": 13.91450259688374,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 34828
    },
    {
      "epoch": 13.914902117459048,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 34829
    },
    {
      "epoch": 13.915301638034359,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 34830
    },
    {
      "epoch": 13.915701158609668,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 34831
    },
    {
      "epoch": 13.916100679184979,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 34832
    },
    {
      "epoch": 13.916500199760288,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 34833
    },
    {
      "epoch": 13.916899720335596,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 34834
    },
    {
      "epoch": 13.917299240910907,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 34835
    },
    {
      "epoch": 13.917698761486216,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 34836
    },
    {
      "epoch": 13.918098282061527,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 34837
    },
    {
      "epoch": 13.918497802636836,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 34838
    },
    {
      "epoch": 13.918897323212146,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 34839
    },
    {
      "epoch": 13.919296843787455,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 34840
    },
    {
      "epoch": 13.919696364362764,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 34841
    },
    {
      "epoch": 13.920095884938075,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 34842
    },
    {
      "epoch": 13.920495405513384,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 34843
    },
    {
      "epoch": 13.920894926088694,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 34844
    },
    {
      "epoch": 13.921294446664003,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 34845
    },
    {
      "epoch": 13.921693967239312,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 34846
    },
    {
      "epoch": 13.922093487814623,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 34847
    },
    {
      "epoch": 13.922493008389932,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 34848
    },
    {
      "epoch": 13.922892528965242,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 34849
    },
    {
      "epoch": 13.923292049540551,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 34850
    },
    {
      "epoch": 13.92369157011586,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 34851
    },
    {
      "epoch": 13.92409109069117,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 34852
    },
    {
      "epoch": 13.92449061126648,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 34853
    },
    {
      "epoch": 13.92489013184179,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 34854
    },
    {
      "epoch": 13.9252896524171,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 34855
    },
    {
      "epoch": 13.92568917299241,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 34856
    },
    {
      "epoch": 13.926088693567719,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 34857
    },
    {
      "epoch": 13.926488214143028,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 34858
    },
    {
      "epoch": 13.926887734718338,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 34859
    },
    {
      "epoch": 13.927287255293647,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 34860
    },
    {
      "epoch": 13.927686775868958,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 34861
    },
    {
      "epoch": 13.928086296444267,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 34862
    },
    {
      "epoch": 13.928485817019576,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 34863
    },
    {
      "epoch": 13.928885337594886,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 34864
    },
    {
      "epoch": 13.929284858170195,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 34865
    },
    {
      "epoch": 13.929684378745506,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 34866
    },
    {
      "epoch": 13.930083899320815,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 34867
    },
    {
      "epoch": 13.930483419896124,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 34868
    },
    {
      "epoch": 13.930882940471434,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 34869
    },
    {
      "epoch": 13.931282461046743,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 34870
    },
    {
      "epoch": 13.931681981622054,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 34871
    },
    {
      "epoch": 13.932081502197363,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 34872
    },
    {
      "epoch": 13.932481022772674,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 34873
    },
    {
      "epoch": 13.932880543347983,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 34874
    },
    {
      "epoch": 13.933280063923291,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 34875
    },
    {
      "epoch": 13.933679584498602,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 34876
    },
    {
      "epoch": 13.934079105073911,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 34877
    },
    {
      "epoch": 13.934478625649222,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 34878
    },
    {
      "epoch": 13.93487814622453,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 34879
    },
    {
      "epoch": 13.93527766679984,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 34880
    },
    {
      "epoch": 13.93567718737515,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 34881
    },
    {
      "epoch": 13.936076707950459,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 34882
    },
    {
      "epoch": 13.93647622852577,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 34883
    },
    {
      "epoch": 13.936875749101079,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 34884
    },
    {
      "epoch": 13.937275269676388,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 34885
    },
    {
      "epoch": 13.937674790251698,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 34886
    },
    {
      "epoch": 13.938074310827007,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 34887
    },
    {
      "epoch": 13.938473831402318,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 34888
    },
    {
      "epoch": 13.938873351977627,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 34889
    },
    {
      "epoch": 13.939272872552937,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 34890
    },
    {
      "epoch": 13.939672393128246,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 34891
    },
    {
      "epoch": 13.940071913703555,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 34892
    },
    {
      "epoch": 13.940471434278866,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 34893
    },
    {
      "epoch": 13.940870954854175,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 34894
    },
    {
      "epoch": 13.941270475429485,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 34895
    },
    {
      "epoch": 13.941669996004794,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 34896
    },
    {
      "epoch": 13.942069516580103,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 34897
    },
    {
      "epoch": 13.942469037155414,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 34898
    },
    {
      "epoch": 13.942868557730723,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 34899
    },
    {
      "epoch": 13.943268078306033,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 34900
    },
    {
      "epoch": 13.943667598881342,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 34901
    },
    {
      "epoch": 13.944067119456651,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 34902
    },
    {
      "epoch": 13.944466640031962,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 34903
    },
    {
      "epoch": 13.94486616060727,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 34904
    },
    {
      "epoch": 13.945265681182581,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 34905
    },
    {
      "epoch": 13.94566520175789,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 34906
    },
    {
      "epoch": 13.946064722333201,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 34907
    },
    {
      "epoch": 13.94646424290851,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 34908
    },
    {
      "epoch": 13.946863763483819,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 34909
    },
    {
      "epoch": 13.94726328405913,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 34910
    },
    {
      "epoch": 13.947662804634438,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 34911
    },
    {
      "epoch": 13.948062325209749,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 34912
    },
    {
      "epoch": 13.948461845785058,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 34913
    },
    {
      "epoch": 13.948861366360367,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 34914
    },
    {
      "epoch": 13.949260886935678,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 34915
    },
    {
      "epoch": 13.949660407510986,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 34916
    },
    {
      "epoch": 13.950059928086297,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 34917
    },
    {
      "epoch": 13.950459448661606,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 34918
    },
    {
      "epoch": 13.950858969236915,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 34919
    },
    {
      "epoch": 13.951258489812226,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 34920
    },
    {
      "epoch": 13.951658010387535,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 34921
    },
    {
      "epoch": 13.952057530962845,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 34922
    },
    {
      "epoch": 13.952457051538154,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 34923
    },
    {
      "epoch": 13.952856572113465,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 34924
    },
    {
      "epoch": 13.953256092688774,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 34925
    },
    {
      "epoch": 13.953655613264083,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 34926
    },
    {
      "epoch": 13.954055133839393,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 34927
    },
    {
      "epoch": 13.954454654414702,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 34928
    },
    {
      "epoch": 13.954854174990013,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 34929
    },
    {
      "epoch": 13.955253695565322,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 34930
    },
    {
      "epoch": 13.95565321614063,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 34931
    },
    {
      "epoch": 13.956052736715941,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 34932
    },
    {
      "epoch": 13.95645225729125,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 34933
    },
    {
      "epoch": 13.95685177786656,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 34934
    },
    {
      "epoch": 13.95725129844187,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 34935
    },
    {
      "epoch": 13.957650819017179,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 34936
    },
    {
      "epoch": 13.95805033959249,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 34937
    },
    {
      "epoch": 13.958449860167798,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 34938
    },
    {
      "epoch": 13.958849380743109,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 34939
    },
    {
      "epoch": 13.959248901318418,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 34940
    },
    {
      "epoch": 13.959648421893728,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 34941
    },
    {
      "epoch": 13.960047942469037,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 34942
    },
    {
      "epoch": 13.960447463044346,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 34943
    },
    {
      "epoch": 13.960846983619657,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 34944
    },
    {
      "epoch": 13.961246504194966,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 34945
    },
    {
      "epoch": 13.961646024770276,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 34946
    },
    {
      "epoch": 13.962045545345585,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 34947
    },
    {
      "epoch": 13.962445065920894,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 34948
    },
    {
      "epoch": 13.962844586496205,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 34949
    },
    {
      "epoch": 13.963244107071514,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 34950
    },
    {
      "epoch": 13.963643627646825,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 34951
    },
    {
      "epoch": 13.964043148222133,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 34952
    },
    {
      "epoch": 13.964442668797442,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 34953
    },
    {
      "epoch": 13.964842189372753,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 34954
    },
    {
      "epoch": 13.965241709948062,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 34955
    },
    {
      "epoch": 13.965641230523373,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 34956
    },
    {
      "epoch": 13.966040751098681,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 34957
    },
    {
      "epoch": 13.96644027167399,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 34958
    },
    {
      "epoch": 13.966839792249301,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 34959
    },
    {
      "epoch": 13.96723931282461,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 34960
    },
    {
      "epoch": 13.96763883339992,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 34961
    },
    {
      "epoch": 13.96803835397523,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 34962
    },
    {
      "epoch": 13.968437874550538,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 34963
    },
    {
      "epoch": 13.968837395125849,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 34964
    },
    {
      "epoch": 13.969236915701158,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 34965
    },
    {
      "epoch": 13.969636436276469,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 34966
    },
    {
      "epoch": 13.970035956851778,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 34967
    },
    {
      "epoch": 13.970435477427088,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 34968
    },
    {
      "epoch": 13.970834998002397,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 34969
    },
    {
      "epoch": 13.971234518577706,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 34970
    },
    {
      "epoch": 13.971634039153017,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 34971
    },
    {
      "epoch": 13.972033559728326,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 34972
    },
    {
      "epoch": 13.972433080303636,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 34973
    },
    {
      "epoch": 13.972832600878945,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 34974
    },
    {
      "epoch": 13.973232121454254,
      "grad_norm": 0.036865234375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 34975
    },
    {
      "epoch": 13.973631642029565,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 34976
    },
    {
      "epoch": 13.974031162604874,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 34977
    },
    {
      "epoch": 13.974430683180184,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 34978
    },
    {
      "epoch": 13.974830203755493,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 34979
    },
    {
      "epoch": 13.975229724330802,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 34980
    },
    {
      "epoch": 13.975629244906113,
      "grad_norm": 0.036865234375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 34981
    },
    {
      "epoch": 13.976028765481422,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 34982
    },
    {
      "epoch": 13.976428286056732,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 34983
    },
    {
      "epoch": 13.976827806632041,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 34984
    },
    {
      "epoch": 13.977227327207352,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 34985
    },
    {
      "epoch": 13.97762684778266,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 34986
    },
    {
      "epoch": 13.97802636835797,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 34987
    },
    {
      "epoch": 13.97842588893328,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 34988
    },
    {
      "epoch": 13.97882540950859,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 34989
    },
    {
      "epoch": 13.9792249300839,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 34990
    },
    {
      "epoch": 13.979624450659209,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 34991
    },
    {
      "epoch": 13.980023971234518,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 34992
    },
    {
      "epoch": 13.980423491809828,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 34993
    },
    {
      "epoch": 13.980823012385137,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 34994
    },
    {
      "epoch": 13.981222532960448,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 34995
    },
    {
      "epoch": 13.981622053535757,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 34996
    },
    {
      "epoch": 13.982021574111066,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 34997
    },
    {
      "epoch": 13.982421094686377,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 34998
    },
    {
      "epoch": 13.982820615261685,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 34999
    },
    {
      "epoch": 13.983220135836996,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 35000
    },
    {
      "epoch": 13.983619656412305,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 35001
    },
    {
      "epoch": 13.984019176987616,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 35002
    },
    {
      "epoch": 13.984418697562925,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 35003
    },
    {
      "epoch": 13.984818218138233,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 35004
    },
    {
      "epoch": 13.985217738713544,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 35005
    },
    {
      "epoch": 13.985617259288853,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 35006
    },
    {
      "epoch": 13.986016779864164,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 35007
    },
    {
      "epoch": 13.986416300439473,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 35008
    },
    {
      "epoch": 13.986815821014781,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 35009
    },
    {
      "epoch": 13.987215341590092,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 35010
    },
    {
      "epoch": 13.987614862165401,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 35011
    },
    {
      "epoch": 13.988014382740712,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 35012
    },
    {
      "epoch": 13.98841390331602,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 35013
    },
    {
      "epoch": 13.98881342389133,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 35014
    },
    {
      "epoch": 13.98921294446664,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 35015
    },
    {
      "epoch": 13.989612465041949,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 35016
    },
    {
      "epoch": 13.99001198561726,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 35017
    },
    {
      "epoch": 13.990411506192569,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 35018
    },
    {
      "epoch": 13.99081102676788,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 35019
    },
    {
      "epoch": 13.991210547343188,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 35020
    },
    {
      "epoch": 13.991610067918497,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 35021
    },
    {
      "epoch": 13.992009588493808,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 35022
    },
    {
      "epoch": 13.992409109069117,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 35023
    },
    {
      "epoch": 13.992808629644427,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 35024
    },
    {
      "epoch": 13.993208150219736,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 35025
    },
    {
      "epoch": 13.993607670795045,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 35026
    },
    {
      "epoch": 13.994007191370356,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 35027
    },
    {
      "epoch": 13.994406711945665,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 35028
    },
    {
      "epoch": 13.994806232520975,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 35029
    },
    {
      "epoch": 13.995205753096284,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 35030
    },
    {
      "epoch": 13.995605273671593,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 35031
    },
    {
      "epoch": 13.996004794246904,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 35032
    },
    {
      "epoch": 13.996404314822213,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 35033
    },
    {
      "epoch": 13.996803835397523,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 35034
    },
    {
      "epoch": 13.997203355972832,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 35035
    },
    {
      "epoch": 13.997602876548143,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 35036
    },
    {
      "epoch": 13.998002397123452,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 35037
    },
    {
      "epoch": 13.99840191769876,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 35038
    },
    {
      "epoch": 13.998801438274072,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 35039
    },
    {
      "epoch": 13.99920095884938,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 35040
    },
    {
      "epoch": 13.999600479424691,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 35041
    },
    {
      "epoch": 14.0,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 35042
    },
    {
      "epoch": 14.000399520575309,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 35043
    },
    {
      "epoch": 14.00079904115062,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 35044
    },
    {
      "epoch": 14.001198561725928,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 35045
    },
    {
      "epoch": 14.00159808230124,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 35046
    },
    {
      "epoch": 14.001997602876548,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 35047
    },
    {
      "epoch": 14.002397123451857,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 35048
    },
    {
      "epoch": 14.002796644027168,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 35049
    },
    {
      "epoch": 14.003196164602477,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 35050
    },
    {
      "epoch": 14.003595685177787,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 35051
    },
    {
      "epoch": 14.003995205753096,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 35052
    },
    {
      "epoch": 14.004394726328407,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 35053
    },
    {
      "epoch": 14.004794246903716,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 35054
    },
    {
      "epoch": 14.005193767479025,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 35055
    },
    {
      "epoch": 14.005593288054335,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 35056
    },
    {
      "epoch": 14.005992808629644,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 35057
    },
    {
      "epoch": 14.006392329204955,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 35058
    },
    {
      "epoch": 14.006791849780264,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 35059
    },
    {
      "epoch": 14.007191370355573,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 35060
    },
    {
      "epoch": 14.007590890930883,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 35061
    },
    {
      "epoch": 14.007990411506192,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 35062
    },
    {
      "epoch": 14.008389932081503,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 35063
    },
    {
      "epoch": 14.008789452656812,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 35064
    },
    {
      "epoch": 14.00918897323212,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 35065
    },
    {
      "epoch": 14.009588493807431,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 35066
    },
    {
      "epoch": 14.00998801438274,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 35067
    },
    {
      "epoch": 14.010387534958051,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 35068
    },
    {
      "epoch": 14.01078705553336,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 35069
    },
    {
      "epoch": 14.01118657610867,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 35070
    },
    {
      "epoch": 14.01158609668398,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 35071
    },
    {
      "epoch": 14.011985617259288,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 35072
    },
    {
      "epoch": 14.012385137834599,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 35073
    },
    {
      "epoch": 14.012784658409908,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 35074
    },
    {
      "epoch": 14.013184178985219,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 35075
    },
    {
      "epoch": 14.013583699560527,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 35076
    },
    {
      "epoch": 14.013983220135836,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 35077
    },
    {
      "epoch": 14.014382740711147,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 35078
    },
    {
      "epoch": 14.014782261286456,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 35079
    },
    {
      "epoch": 14.015181781861767,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 35080
    },
    {
      "epoch": 14.015581302437075,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 35081
    },
    {
      "epoch": 14.015980823012384,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 35082
    },
    {
      "epoch": 14.016380343587695,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 35083
    },
    {
      "epoch": 14.016779864163004,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 35084
    },
    {
      "epoch": 14.017179384738315,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 35085
    },
    {
      "epoch": 14.017578905313623,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 35086
    },
    {
      "epoch": 14.017978425888934,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 35087
    },
    {
      "epoch": 14.018377946464243,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 35088
    },
    {
      "epoch": 14.018777467039552,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 35089
    },
    {
      "epoch": 14.019176987614863,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 35090
    },
    {
      "epoch": 14.019576508190172,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 35091
    },
    {
      "epoch": 14.019976028765482,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 35092
    },
    {
      "epoch": 14.020375549340791,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 35093
    },
    {
      "epoch": 14.0207750699161,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 35094
    },
    {
      "epoch": 14.02117459049141,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 35095
    },
    {
      "epoch": 14.02157411106672,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.046,
      "step": 35096
    },
    {
      "epoch": 14.02197363164203,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 35097
    },
    {
      "epoch": 14.02237315221734,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 35098
    },
    {
      "epoch": 14.022772672792648,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 35099
    },
    {
      "epoch": 14.023172193367959,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 35100
    },
    {
      "epoch": 14.023571713943268,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 35101
    },
    {
      "epoch": 14.023971234518578,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 35102
    },
    {
      "epoch": 14.024370755093887,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 35103
    },
    {
      "epoch": 14.024770275669196,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 35104
    },
    {
      "epoch": 14.025169796244507,
      "grad_norm": 0.036865234375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 35105
    },
    {
      "epoch": 14.025569316819816,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 35106
    },
    {
      "epoch": 14.025968837395126,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 35107
    },
    {
      "epoch": 14.026368357970435,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 35108
    },
    {
      "epoch": 14.026767878545746,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 35109
    },
    {
      "epoch": 14.027167399121055,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 35110
    },
    {
      "epoch": 14.027566919696364,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 35111
    },
    {
      "epoch": 14.027966440271674,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 35112
    },
    {
      "epoch": 14.028365960846983,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 35113
    },
    {
      "epoch": 14.028765481422294,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 35114
    },
    {
      "epoch": 14.029165001997603,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 35115
    },
    {
      "epoch": 14.029564522572912,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 35116
    },
    {
      "epoch": 14.029964043148222,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 35117
    },
    {
      "epoch": 14.030363563723531,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 35118
    },
    {
      "epoch": 14.030763084298842,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 35119
    },
    {
      "epoch": 14.031162604874151,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 35120
    },
    {
      "epoch": 14.03156212544946,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 35121
    },
    {
      "epoch": 14.03196164602477,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 35122
    },
    {
      "epoch": 14.03236116660008,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 35123
    },
    {
      "epoch": 14.03276068717539,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 35124
    },
    {
      "epoch": 14.033160207750699,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 35125
    },
    {
      "epoch": 14.03355972832601,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 35126
    },
    {
      "epoch": 14.033959248901319,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 35127
    },
    {
      "epoch": 14.034358769476627,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 35128
    },
    {
      "epoch": 14.034758290051938,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 35129
    },
    {
      "epoch": 14.035157810627247,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 35130
    },
    {
      "epoch": 14.035557331202558,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 35131
    },
    {
      "epoch": 14.035956851777867,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 35132
    },
    {
      "epoch": 14.036356372353175,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 35133
    },
    {
      "epoch": 14.036755892928486,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 35134
    },
    {
      "epoch": 14.037155413503795,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 35135
    },
    {
      "epoch": 14.037554934079106,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 35136
    },
    {
      "epoch": 14.037954454654415,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 35137
    },
    {
      "epoch": 14.038353975229724,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 35138
    },
    {
      "epoch": 14.038753495805034,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 35139
    },
    {
      "epoch": 14.039153016380343,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 35140
    },
    {
      "epoch": 14.039552536955654,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 35141
    },
    {
      "epoch": 14.039952057530963,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 35142
    },
    {
      "epoch": 14.040351578106273,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 35143
    },
    {
      "epoch": 14.040751098681582,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 35144
    },
    {
      "epoch": 14.041150619256891,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 35145
    },
    {
      "epoch": 14.041550139832202,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 35146
    },
    {
      "epoch": 14.04194966040751,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 35147
    },
    {
      "epoch": 14.042349180982821,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 35148
    },
    {
      "epoch": 14.04274870155813,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 35149
    },
    {
      "epoch": 14.04314822213344,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 35150
    },
    {
      "epoch": 14.04354774270875,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 35151
    },
    {
      "epoch": 14.043947263284059,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0116,
      "step": 35152
    },
    {
      "epoch": 14.04434678385937,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 35153
    },
    {
      "epoch": 14.044746304434678,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 35154
    },
    {
      "epoch": 14.045145825009987,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 35155
    },
    {
      "epoch": 14.045545345585298,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 35156
    },
    {
      "epoch": 14.045944866160607,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 35157
    },
    {
      "epoch": 14.046344386735917,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 35158
    },
    {
      "epoch": 14.046743907311226,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 35159
    },
    {
      "epoch": 14.047143427886537,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 35160
    },
    {
      "epoch": 14.047542948461846,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 35161
    },
    {
      "epoch": 14.047942469037155,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 35162
    },
    {
      "epoch": 14.048341989612465,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 35163
    },
    {
      "epoch": 14.048741510187774,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 35164
    },
    {
      "epoch": 14.049141030763085,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 35165
    },
    {
      "epoch": 14.049540551338394,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 35166
    },
    {
      "epoch": 14.049940071913703,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 35167
    },
    {
      "epoch": 14.050339592489014,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 35168
    },
    {
      "epoch": 14.050739113064322,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 35169
    },
    {
      "epoch": 14.051138633639633,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 35170
    },
    {
      "epoch": 14.051538154214942,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 35171
    },
    {
      "epoch": 14.051937674790251,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 35172
    },
    {
      "epoch": 14.052337195365562,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 35173
    },
    {
      "epoch": 14.05273671594087,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0092,
      "step": 35174
    },
    {
      "epoch": 14.053136236516181,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 35175
    },
    {
      "epoch": 14.05353575709149,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 35176
    },
    {
      "epoch": 14.053935277666799,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 35177
    },
    {
      "epoch": 14.05433479824211,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 35178
    },
    {
      "epoch": 14.054734318817419,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 35179
    },
    {
      "epoch": 14.05513383939273,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 35180
    },
    {
      "epoch": 14.055533359968038,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 35181
    },
    {
      "epoch": 14.055932880543349,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 35182
    },
    {
      "epoch": 14.056332401118658,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 35183
    },
    {
      "epoch": 14.056731921693967,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 35184
    },
    {
      "epoch": 14.057131442269277,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 35185
    },
    {
      "epoch": 14.057530962844586,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 35186
    },
    {
      "epoch": 14.057930483419897,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 35187
    },
    {
      "epoch": 14.058330003995206,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 35188
    },
    {
      "epoch": 14.058729524570515,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 35189
    },
    {
      "epoch": 14.059129045145825,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 35190
    },
    {
      "epoch": 14.059528565721134,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 35191
    },
    {
      "epoch": 14.059928086296445,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 35192
    },
    {
      "epoch": 14.060327606871754,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 35193
    },
    {
      "epoch": 14.060727127447063,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 35194
    },
    {
      "epoch": 14.061126648022373,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 35195
    },
    {
      "epoch": 14.061526168597682,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 35196
    },
    {
      "epoch": 14.061925689172993,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 35197
    },
    {
      "epoch": 14.062325209748302,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 35198
    },
    {
      "epoch": 14.062724730323612,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 35199
    },
    {
      "epoch": 14.063124250898921,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 35200
    },
    {
      "epoch": 14.06352377147423,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 35201
    },
    {
      "epoch": 14.063923292049541,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 35202
    },
    {
      "epoch": 14.06432281262485,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 35203
    },
    {
      "epoch": 14.06472233320016,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 35204
    },
    {
      "epoch": 14.06512185377547,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 35205
    },
    {
      "epoch": 14.065521374350778,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 35206
    },
    {
      "epoch": 14.065920894926089,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 35207
    },
    {
      "epoch": 14.066320415501398,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 35208
    },
    {
      "epoch": 14.066719936076709,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 35209
    },
    {
      "epoch": 14.067119456652017,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 35210
    },
    {
      "epoch": 14.067518977227326,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 35211
    },
    {
      "epoch": 14.067918497802637,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 35212
    },
    {
      "epoch": 14.068318018377946,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 35213
    },
    {
      "epoch": 14.068717538953257,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 35214
    },
    {
      "epoch": 14.069117059528566,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 35215
    },
    {
      "epoch": 14.069516580103876,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 35216
    },
    {
      "epoch": 14.069916100679185,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 35217
    },
    {
      "epoch": 14.070315621254494,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 35218
    },
    {
      "epoch": 14.070715141829805,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 35219
    },
    {
      "epoch": 14.071114662405114,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 35220
    },
    {
      "epoch": 14.071514182980424,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 35221
    },
    {
      "epoch": 14.071913703555733,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 35222
    },
    {
      "epoch": 14.072313224131042,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 35223
    },
    {
      "epoch": 14.072712744706353,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 35224
    },
    {
      "epoch": 14.073112265281662,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 35225
    },
    {
      "epoch": 14.073511785856972,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 35226
    },
    {
      "epoch": 14.073911306432281,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 35227
    },
    {
      "epoch": 14.07431082700759,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 35228
    },
    {
      "epoch": 14.0747103475829,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 35229
    },
    {
      "epoch": 14.07510986815821,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 35230
    },
    {
      "epoch": 14.07550938873352,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 35231
    },
    {
      "epoch": 14.07590890930883,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 35232
    },
    {
      "epoch": 14.07630842988414,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 35233
    },
    {
      "epoch": 14.076707950459449,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 35234
    },
    {
      "epoch": 14.077107471034758,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 35235
    },
    {
      "epoch": 14.077506991610068,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 35236
    },
    {
      "epoch": 14.077906512185377,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 35237
    },
    {
      "epoch": 14.078306032760688,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 35238
    },
    {
      "epoch": 14.078705553335997,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 35239
    },
    {
      "epoch": 14.079105073911306,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 35240
    },
    {
      "epoch": 14.079504594486616,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 35241
    },
    {
      "epoch": 14.079904115061925,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 35242
    },
    {
      "epoch": 14.080303635637236,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 35243
    },
    {
      "epoch": 14.080703156212545,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 35244
    },
    {
      "epoch": 14.081102676787854,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 35245
    },
    {
      "epoch": 14.081502197363164,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 35246
    },
    {
      "epoch": 14.081901717938473,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 35247
    },
    {
      "epoch": 14.082301238513784,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 35248
    },
    {
      "epoch": 14.082700759089093,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 35249
    },
    {
      "epoch": 14.083100279664404,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 35250
    },
    {
      "epoch": 14.083499800239712,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 35251
    },
    {
      "epoch": 14.083899320815021,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 35252
    },
    {
      "epoch": 14.084298841390332,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 35253
    },
    {
      "epoch": 14.084698361965641,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 35254
    },
    {
      "epoch": 14.085097882540952,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 35255
    },
    {
      "epoch": 14.08549740311626,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 35256
    },
    {
      "epoch": 14.08589692369157,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0116,
      "step": 35257
    },
    {
      "epoch": 14.08629644426688,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 35258
    },
    {
      "epoch": 14.086695964842189,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 35259
    },
    {
      "epoch": 14.0870954854175,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 35260
    },
    {
      "epoch": 14.087495005992809,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 35261
    },
    {
      "epoch": 14.087894526568117,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 35262
    },
    {
      "epoch": 14.088294047143428,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 35263
    },
    {
      "epoch": 14.088693567718737,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0118,
      "step": 35264
    },
    {
      "epoch": 14.089093088294048,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 35265
    },
    {
      "epoch": 14.089492608869357,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 35266
    },
    {
      "epoch": 14.089892129444667,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 35267
    },
    {
      "epoch": 14.090291650019976,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 35268
    },
    {
      "epoch": 14.090691170595285,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 35269
    },
    {
      "epoch": 14.091090691170596,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 35270
    },
    {
      "epoch": 14.091490211745905,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 35271
    },
    {
      "epoch": 14.091889732321215,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 35272
    },
    {
      "epoch": 14.092289252896524,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 35273
    },
    {
      "epoch": 14.092688773471833,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 35274
    },
    {
      "epoch": 14.093088294047144,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 35275
    },
    {
      "epoch": 14.093487814622453,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 35276
    },
    {
      "epoch": 14.093887335197763,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 35277
    },
    {
      "epoch": 14.094286855773072,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 35278
    },
    {
      "epoch": 14.094686376348381,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 35279
    },
    {
      "epoch": 14.095085896923692,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 35280
    },
    {
      "epoch": 14.095485417499,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 35281
    },
    {
      "epoch": 14.095884938074311,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 35282
    },
    {
      "epoch": 14.09628445864962,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 35283
    },
    {
      "epoch": 14.09668397922493,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 35284
    },
    {
      "epoch": 14.09708349980024,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 35285
    },
    {
      "epoch": 14.097483020375549,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 35286
    },
    {
      "epoch": 14.09788254095086,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 35287
    },
    {
      "epoch": 14.098282061526168,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 35288
    },
    {
      "epoch": 14.098681582101479,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 35289
    },
    {
      "epoch": 14.099081102676788,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 35290
    },
    {
      "epoch": 14.099480623252097,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 35291
    },
    {
      "epoch": 14.099880143827408,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 35292
    },
    {
      "epoch": 14.100279664402716,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 35293
    },
    {
      "epoch": 14.100679184978027,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 35294
    },
    {
      "epoch": 14.101078705553336,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 35295
    },
    {
      "epoch": 14.101478226128645,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 35296
    },
    {
      "epoch": 14.101877746703956,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 35297
    },
    {
      "epoch": 14.102277267279264,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 35298
    },
    {
      "epoch": 14.102676787854575,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 35299
    },
    {
      "epoch": 14.103076308429884,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 35300
    },
    {
      "epoch": 14.103475829005193,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 35301
    },
    {
      "epoch": 14.103875349580504,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 35302
    },
    {
      "epoch": 14.104274870155812,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 35303
    },
    {
      "epoch": 14.104674390731123,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 35304
    },
    {
      "epoch": 14.105073911306432,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 35305
    },
    {
      "epoch": 14.105473431881743,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 35306
    },
    {
      "epoch": 14.105872952457052,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 35307
    },
    {
      "epoch": 14.10627247303236,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.012,
      "step": 35308
    },
    {
      "epoch": 14.106671993607671,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 35309
    },
    {
      "epoch": 14.10707151418298,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 35310
    },
    {
      "epoch": 14.10747103475829,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 35311
    },
    {
      "epoch": 14.1078705553336,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 35312
    },
    {
      "epoch": 14.108270075908909,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 35313
    },
    {
      "epoch": 14.10866959648422,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 35314
    },
    {
      "epoch": 14.109069117059528,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 35315
    },
    {
      "epoch": 14.109468637634839,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 35316
    },
    {
      "epoch": 14.109868158210148,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 35317
    },
    {
      "epoch": 14.110267678785457,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 35318
    },
    {
      "epoch": 14.110667199360767,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 35319
    },
    {
      "epoch": 14.111066719936076,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 35320
    },
    {
      "epoch": 14.111466240511387,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 35321
    },
    {
      "epoch": 14.111865761086696,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 35322
    },
    {
      "epoch": 14.112265281662006,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 35323
    },
    {
      "epoch": 14.112664802237315,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 35324
    },
    {
      "epoch": 14.113064322812624,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 35325
    },
    {
      "epoch": 14.113463843387935,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 35326
    },
    {
      "epoch": 14.113863363963244,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 35327
    },
    {
      "epoch": 14.114262884538554,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 35328
    },
    {
      "epoch": 14.114662405113863,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 35329
    },
    {
      "epoch": 14.115061925689172,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 35330
    },
    {
      "epoch": 14.115461446264483,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 35331
    },
    {
      "epoch": 14.115860966839792,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 35332
    },
    {
      "epoch": 14.116260487415103,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 35333
    },
    {
      "epoch": 14.116660007990411,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 35334
    },
    {
      "epoch": 14.11705952856572,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 35335
    },
    {
      "epoch": 14.117459049141031,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 35336
    },
    {
      "epoch": 14.11785856971634,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 35337
    },
    {
      "epoch": 14.11825809029165,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 35338
    },
    {
      "epoch": 14.11865761086696,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 35339
    },
    {
      "epoch": 14.11905713144227,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 35340
    },
    {
      "epoch": 14.119456652017579,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 35341
    },
    {
      "epoch": 14.119856172592888,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 35342
    },
    {
      "epoch": 14.120255693168199,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 35343
    },
    {
      "epoch": 14.120655213743508,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 35344
    },
    {
      "epoch": 14.121054734318818,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 35345
    },
    {
      "epoch": 14.121454254894127,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 35346
    },
    {
      "epoch": 14.121853775469436,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 35347
    },
    {
      "epoch": 14.122253296044747,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 35348
    },
    {
      "epoch": 14.122652816620056,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0101,
      "step": 35349
    },
    {
      "epoch": 14.123052337195366,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 35350
    },
    {
      "epoch": 14.123451857770675,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 35351
    },
    {
      "epoch": 14.123851378345984,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 35352
    },
    {
      "epoch": 14.124250898921295,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 35353
    },
    {
      "epoch": 14.124650419496604,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 35354
    },
    {
      "epoch": 14.125049940071914,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 35355
    },
    {
      "epoch": 14.125449460647223,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 35356
    },
    {
      "epoch": 14.125848981222532,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 35357
    },
    {
      "epoch": 14.126248501797843,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 35358
    },
    {
      "epoch": 14.126648022373152,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 35359
    },
    {
      "epoch": 14.127047542948462,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 35360
    },
    {
      "epoch": 14.127447063523771,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 35361
    },
    {
      "epoch": 14.127846584099082,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 35362
    },
    {
      "epoch": 14.12824610467439,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 35363
    },
    {
      "epoch": 14.1286456252497,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 35364
    },
    {
      "epoch": 14.12904514582501,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 35365
    },
    {
      "epoch": 14.12944466640032,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 35366
    },
    {
      "epoch": 14.12984418697563,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 35367
    },
    {
      "epoch": 14.130243707550939,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 35368
    },
    {
      "epoch": 14.130643228126248,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 35369
    },
    {
      "epoch": 14.131042748701558,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 35370
    },
    {
      "epoch": 14.131442269276867,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 35371
    },
    {
      "epoch": 14.131841789852178,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 35372
    },
    {
      "epoch": 14.132241310427487,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 35373
    },
    {
      "epoch": 14.132640831002796,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 35374
    },
    {
      "epoch": 14.133040351578106,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 35375
    },
    {
      "epoch": 14.133439872153415,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 35376
    },
    {
      "epoch": 14.133839392728726,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 35377
    },
    {
      "epoch": 14.134238913304035,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 35378
    },
    {
      "epoch": 14.134638433879346,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 35379
    },
    {
      "epoch": 14.135037954454654,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 35380
    },
    {
      "epoch": 14.135437475029963,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 35381
    },
    {
      "epoch": 14.135836995605274,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 35382
    },
    {
      "epoch": 14.136236516180583,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 35383
    },
    {
      "epoch": 14.136636036755894,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 35384
    },
    {
      "epoch": 14.137035557331203,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 35385
    },
    {
      "epoch": 14.137435077906511,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 35386
    },
    {
      "epoch": 14.137834598481822,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 35387
    },
    {
      "epoch": 14.138234119057131,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 35388
    },
    {
      "epoch": 14.138633639632442,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 35389
    },
    {
      "epoch": 14.13903316020775,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 35390
    },
    {
      "epoch": 14.13943268078306,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 35391
    },
    {
      "epoch": 14.13983220135837,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 35392
    },
    {
      "epoch": 14.140231721933679,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 35393
    },
    {
      "epoch": 14.14063124250899,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 35394
    },
    {
      "epoch": 14.141030763084299,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 35395
    },
    {
      "epoch": 14.14143028365961,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 35396
    },
    {
      "epoch": 14.141829804234918,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 35397
    },
    {
      "epoch": 14.142229324810227,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 35398
    },
    {
      "epoch": 14.142628845385538,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 35399
    },
    {
      "epoch": 14.143028365960847,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 35400
    },
    {
      "epoch": 14.143427886536157,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 35401
    },
    {
      "epoch": 14.143827407111466,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 35402
    },
    {
      "epoch": 14.144226927686775,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 35403
    },
    {
      "epoch": 14.144626448262086,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 35404
    },
    {
      "epoch": 14.145025968837395,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 35405
    },
    {
      "epoch": 14.145425489412705,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 35406
    },
    {
      "epoch": 14.145825009988014,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 35407
    },
    {
      "epoch": 14.146224530563323,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 35408
    },
    {
      "epoch": 14.146624051138634,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 35409
    },
    {
      "epoch": 14.147023571713943,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 35410
    },
    {
      "epoch": 14.147423092289253,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 35411
    },
    {
      "epoch": 14.147822612864562,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 35412
    },
    {
      "epoch": 14.148222133439873,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 35413
    },
    {
      "epoch": 14.148621654015182,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 35414
    },
    {
      "epoch": 14.14902117459049,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 35415
    },
    {
      "epoch": 14.149420695165801,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 35416
    },
    {
      "epoch": 14.14982021574111,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 35417
    },
    {
      "epoch": 14.150219736316421,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 35418
    },
    {
      "epoch": 14.15061925689173,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 35419
    },
    {
      "epoch": 14.151018777467039,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 35420
    },
    {
      "epoch": 14.15141829804235,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 35421
    },
    {
      "epoch": 14.151817818617658,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 35422
    },
    {
      "epoch": 14.152217339192969,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 35423
    },
    {
      "epoch": 14.152616859768278,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 35424
    },
    {
      "epoch": 14.153016380343587,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 35425
    },
    {
      "epoch": 14.153415900918898,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 35426
    },
    {
      "epoch": 14.153815421494206,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 35427
    },
    {
      "epoch": 14.154214942069517,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 35428
    },
    {
      "epoch": 14.154614462644826,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 35429
    },
    {
      "epoch": 14.155013983220137,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 35430
    },
    {
      "epoch": 14.155413503795446,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 35431
    },
    {
      "epoch": 14.155813024370755,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 35432
    },
    {
      "epoch": 14.156212544946065,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 35433
    },
    {
      "epoch": 14.156612065521374,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 35434
    },
    {
      "epoch": 14.157011586096685,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 35435
    },
    {
      "epoch": 14.157411106671994,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 35436
    },
    {
      "epoch": 14.157810627247303,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 35437
    },
    {
      "epoch": 14.158210147822613,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 35438
    },
    {
      "epoch": 14.158609668397922,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 35439
    },
    {
      "epoch": 14.159009188973233,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 35440
    },
    {
      "epoch": 14.159408709548542,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 35441
    },
    {
      "epoch": 14.15980823012385,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 35442
    },
    {
      "epoch": 14.160207750699161,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 35443
    },
    {
      "epoch": 14.16060727127447,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 35444
    },
    {
      "epoch": 14.16100679184978,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 35445
    },
    {
      "epoch": 14.16140631242509,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 35446
    },
    {
      "epoch": 14.1618058330004,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 35447
    },
    {
      "epoch": 14.16220535357571,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 35448
    },
    {
      "epoch": 14.162604874151018,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 35449
    },
    {
      "epoch": 14.163004394726329,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 35450
    },
    {
      "epoch": 14.163403915301638,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 35451
    },
    {
      "epoch": 14.163803435876948,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 35452
    },
    {
      "epoch": 14.164202956452257,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 35453
    },
    {
      "epoch": 14.164602477027566,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 35454
    },
    {
      "epoch": 14.165001997602877,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 35455
    },
    {
      "epoch": 14.165401518178186,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 35456
    },
    {
      "epoch": 14.165801038753496,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 35457
    },
    {
      "epoch": 14.166200559328805,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 35458
    },
    {
      "epoch": 14.166600079904114,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 35459
    },
    {
      "epoch": 14.166999600479425,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 35460
    },
    {
      "epoch": 14.167399121054734,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 35461
    },
    {
      "epoch": 14.167798641630045,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 35462
    },
    {
      "epoch": 14.168198162205353,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 35463
    },
    {
      "epoch": 14.168597682780662,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 35464
    },
    {
      "epoch": 14.168997203355973,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 35465
    },
    {
      "epoch": 14.169396723931282,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 35466
    },
    {
      "epoch": 14.169796244506593,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 35467
    },
    {
      "epoch": 14.170195765081901,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 35468
    },
    {
      "epoch": 14.170595285657212,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 35469
    },
    {
      "epoch": 14.170994806232521,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 35470
    },
    {
      "epoch": 14.17139432680783,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 35471
    },
    {
      "epoch": 14.17179384738314,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 35472
    },
    {
      "epoch": 14.17219336795845,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 35473
    },
    {
      "epoch": 14.17259288853376,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 35474
    },
    {
      "epoch": 14.172992409109069,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 35475
    },
    {
      "epoch": 14.173391929684378,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 35476
    },
    {
      "epoch": 14.173791450259689,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 35477
    },
    {
      "epoch": 14.174190970834998,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 35478
    },
    {
      "epoch": 14.174590491410308,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 35479
    },
    {
      "epoch": 14.174990011985617,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 35480
    },
    {
      "epoch": 14.175389532560926,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 35481
    },
    {
      "epoch": 14.175789053136237,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 35482
    },
    {
      "epoch": 14.176188573711546,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 35483
    },
    {
      "epoch": 14.176588094286856,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 35484
    },
    {
      "epoch": 14.176987614862165,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 35485
    },
    {
      "epoch": 14.177387135437476,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 35486
    },
    {
      "epoch": 14.177786656012785,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 35487
    },
    {
      "epoch": 14.178186176588094,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 35488
    },
    {
      "epoch": 14.178585697163404,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 35489
    },
    {
      "epoch": 14.178985217738713,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 35490
    },
    {
      "epoch": 14.179384738314024,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 35491
    },
    {
      "epoch": 14.179784258889333,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 35492
    },
    {
      "epoch": 14.180183779464642,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 35493
    },
    {
      "epoch": 14.180583300039952,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 35494
    },
    {
      "epoch": 14.180982820615261,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 35495
    },
    {
      "epoch": 14.181382341190572,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 35496
    },
    {
      "epoch": 14.18178186176588,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 35497
    },
    {
      "epoch": 14.18218138234119,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 35498
    },
    {
      "epoch": 14.1825809029165,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 35499
    },
    {
      "epoch": 14.18298042349181,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 35500
    },
    {
      "epoch": 14.18337994406712,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 35501
    },
    {
      "epoch": 14.183779464642429,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 35502
    },
    {
      "epoch": 14.18417898521774,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 35503
    },
    {
      "epoch": 14.184578505793048,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 35504
    },
    {
      "epoch": 14.184978026368357,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 35505
    },
    {
      "epoch": 14.185377546943668,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 35506
    },
    {
      "epoch": 14.185777067518977,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 35507
    },
    {
      "epoch": 14.186176588094288,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 35508
    },
    {
      "epoch": 14.186576108669597,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 35509
    },
    {
      "epoch": 14.186975629244905,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 35510
    },
    {
      "epoch": 14.187375149820216,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 35511
    },
    {
      "epoch": 14.187774670395525,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 35512
    },
    {
      "epoch": 14.188174190970836,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 35513
    },
    {
      "epoch": 14.188573711546145,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 35514
    },
    {
      "epoch": 14.188973232121453,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 35515
    },
    {
      "epoch": 14.189372752696764,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 35516
    },
    {
      "epoch": 14.189772273272073,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 35517
    },
    {
      "epoch": 14.190171793847384,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 35518
    },
    {
      "epoch": 14.190571314422693,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 35519
    },
    {
      "epoch": 14.190970834998003,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 35520
    },
    {
      "epoch": 14.191370355573312,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 35521
    },
    {
      "epoch": 14.191769876148621,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 35522
    },
    {
      "epoch": 14.192169396723932,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 35523
    },
    {
      "epoch": 14.19256891729924,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 35524
    },
    {
      "epoch": 14.192968437874551,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 35525
    },
    {
      "epoch": 14.19336795844986,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 35526
    },
    {
      "epoch": 14.193767479025169,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 35527
    },
    {
      "epoch": 14.19416699960048,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 35528
    },
    {
      "epoch": 14.194566520175789,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 35529
    },
    {
      "epoch": 14.1949660407511,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 35530
    },
    {
      "epoch": 14.195365561326408,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 35531
    },
    {
      "epoch": 14.195765081901717,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 35532
    },
    {
      "epoch": 14.196164602477028,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 35533
    },
    {
      "epoch": 14.196564123052337,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 35534
    },
    {
      "epoch": 14.196963643627647,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 35535
    },
    {
      "epoch": 14.197363164202956,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 35536
    },
    {
      "epoch": 14.197762684778265,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0046,
      "step": 35537
    },
    {
      "epoch": 14.198162205353576,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 35538
    },
    {
      "epoch": 14.198561725928885,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 35539
    },
    {
      "epoch": 14.198961246504195,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.012,
      "step": 35540
    },
    {
      "epoch": 14.199360767079504,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 35541
    },
    {
      "epoch": 14.199760287654815,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 35542
    },
    {
      "epoch": 14.200159808230124,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 35543
    },
    {
      "epoch": 14.200559328805433,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 35544
    },
    {
      "epoch": 14.200958849380743,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 35545
    },
    {
      "epoch": 14.201358369956052,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 35546
    },
    {
      "epoch": 14.201757890531363,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 35547
    },
    {
      "epoch": 14.202157411106672,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 35548
    },
    {
      "epoch": 14.20255693168198,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 35549
    },
    {
      "epoch": 14.202956452257292,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 35550
    },
    {
      "epoch": 14.2033559728326,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 35551
    },
    {
      "epoch": 14.203755493407911,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 35552
    },
    {
      "epoch": 14.20415501398322,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 35553
    },
    {
      "epoch": 14.204554534558529,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 35554
    },
    {
      "epoch": 14.20495405513384,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 35555
    },
    {
      "epoch": 14.205353575709148,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 35556
    },
    {
      "epoch": 14.20575309628446,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 35557
    },
    {
      "epoch": 14.206152616859768,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 35558
    },
    {
      "epoch": 14.206552137435079,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 35559
    },
    {
      "epoch": 14.206951658010388,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 35560
    },
    {
      "epoch": 14.207351178585697,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 35561
    },
    {
      "epoch": 14.207750699161007,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 35562
    },
    {
      "epoch": 14.208150219736316,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 35563
    },
    {
      "epoch": 14.208549740311627,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 35564
    },
    {
      "epoch": 14.208949260886936,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 35565
    },
    {
      "epoch": 14.209348781462245,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 35566
    },
    {
      "epoch": 14.209748302037555,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 35567
    },
    {
      "epoch": 14.210147822612864,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 35568
    },
    {
      "epoch": 14.210547343188175,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 35569
    },
    {
      "epoch": 14.210946863763484,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 35570
    },
    {
      "epoch": 14.211346384338793,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 35571
    },
    {
      "epoch": 14.211745904914103,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 35572
    },
    {
      "epoch": 14.212145425489412,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 35573
    },
    {
      "epoch": 14.212544946064723,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 35574
    },
    {
      "epoch": 14.212944466640032,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 35575
    },
    {
      "epoch": 14.213343987215342,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 35576
    },
    {
      "epoch": 14.213743507790651,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 35577
    },
    {
      "epoch": 14.21414302836596,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 35578
    },
    {
      "epoch": 14.214542548941271,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 35579
    },
    {
      "epoch": 14.21494206951658,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 35580
    },
    {
      "epoch": 14.21534159009189,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 35581
    },
    {
      "epoch": 14.2157411106672,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 35582
    },
    {
      "epoch": 14.216140631242508,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 35583
    },
    {
      "epoch": 14.216540151817819,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 35584
    },
    {
      "epoch": 14.216939672393128,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 35585
    },
    {
      "epoch": 14.217339192968439,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 35586
    },
    {
      "epoch": 14.217738713543747,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 35587
    },
    {
      "epoch": 14.218138234119056,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 35588
    },
    {
      "epoch": 14.218537754694367,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 35589
    },
    {
      "epoch": 14.218937275269676,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 35590
    },
    {
      "epoch": 14.219336795844987,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 35591
    },
    {
      "epoch": 14.219736316420295,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 35592
    },
    {
      "epoch": 14.220135836995606,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 35593
    },
    {
      "epoch": 14.220535357570915,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0127,
      "step": 35594
    },
    {
      "epoch": 14.220934878146224,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 35595
    },
    {
      "epoch": 14.221334398721535,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 35596
    },
    {
      "epoch": 14.221733919296843,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 35597
    },
    {
      "epoch": 14.222133439872154,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 35598
    },
    {
      "epoch": 14.222532960447463,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 35599
    },
    {
      "epoch": 14.222932481022772,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 35600
    },
    {
      "epoch": 14.223332001598083,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 35601
    },
    {
      "epoch": 14.223731522173392,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 35602
    },
    {
      "epoch": 14.224131042748702,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 35603
    },
    {
      "epoch": 14.224530563324011,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 35604
    },
    {
      "epoch": 14.22493008389932,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 35605
    },
    {
      "epoch": 14.22532960447463,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 35606
    },
    {
      "epoch": 14.22572912504994,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 35607
    },
    {
      "epoch": 14.22612864562525,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 35608
    },
    {
      "epoch": 14.22652816620056,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 35609
    },
    {
      "epoch": 14.22692768677587,
      "grad_norm": 0.036865234375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 35610
    },
    {
      "epoch": 14.227327207351179,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 35611
    },
    {
      "epoch": 14.227726727926488,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 35612
    },
    {
      "epoch": 14.228126248501798,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 35613
    },
    {
      "epoch": 14.228525769077107,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 35614
    },
    {
      "epoch": 14.228925289652418,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 35615
    },
    {
      "epoch": 14.229324810227727,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 35616
    },
    {
      "epoch": 14.229724330803036,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 35617
    },
    {
      "epoch": 14.230123851378346,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 35618
    },
    {
      "epoch": 14.230523371953655,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 35619
    },
    {
      "epoch": 14.230922892528966,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 35620
    },
    {
      "epoch": 14.231322413104275,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 35621
    },
    {
      "epoch": 14.231721933679584,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 35622
    },
    {
      "epoch": 14.232121454254894,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 35623
    },
    {
      "epoch": 14.232520974830203,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 35624
    },
    {
      "epoch": 14.232920495405514,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 35625
    },
    {
      "epoch": 14.233320015980823,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 35626
    },
    {
      "epoch": 14.233719536556134,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 35627
    },
    {
      "epoch": 14.234119057131442,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 35628
    },
    {
      "epoch": 14.234518577706751,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 35629
    },
    {
      "epoch": 14.234918098282062,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 35630
    },
    {
      "epoch": 14.235317618857371,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 35631
    },
    {
      "epoch": 14.235717139432682,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 35632
    },
    {
      "epoch": 14.23611666000799,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 35633
    },
    {
      "epoch": 14.2365161805833,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 35634
    },
    {
      "epoch": 14.23691570115861,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 35635
    },
    {
      "epoch": 14.237315221733919,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 35636
    },
    {
      "epoch": 14.23771474230923,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 35637
    },
    {
      "epoch": 14.238114262884539,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 35638
    },
    {
      "epoch": 14.238513783459847,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 35639
    },
    {
      "epoch": 14.238913304035158,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 35640
    },
    {
      "epoch": 14.239312824610467,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 35641
    },
    {
      "epoch": 14.239712345185778,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 35642
    },
    {
      "epoch": 14.240111865761087,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 35643
    },
    {
      "epoch": 14.240511386336395,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 35644
    },
    {
      "epoch": 14.240910906911706,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 35645
    },
    {
      "epoch": 14.241310427487015,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 35646
    },
    {
      "epoch": 14.241709948062326,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 35647
    },
    {
      "epoch": 14.242109468637635,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 35648
    },
    {
      "epoch": 14.242508989212945,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 35649
    },
    {
      "epoch": 14.242908509788254,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 35650
    },
    {
      "epoch": 14.243308030363563,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 35651
    },
    {
      "epoch": 14.243707550938874,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0085,
      "step": 35652
    },
    {
      "epoch": 14.244107071514183,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 35653
    },
    {
      "epoch": 14.244506592089493,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 35654
    },
    {
      "epoch": 14.244906112664802,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 35655
    },
    {
      "epoch": 14.245305633240111,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 35656
    },
    {
      "epoch": 14.245705153815422,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 35657
    },
    {
      "epoch": 14.24610467439073,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 35658
    },
    {
      "epoch": 14.246504194966041,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 35659
    },
    {
      "epoch": 14.24690371554135,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 35660
    },
    {
      "epoch": 14.24730323611666,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 35661
    },
    {
      "epoch": 14.24770275669197,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 35662
    },
    {
      "epoch": 14.248102277267279,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 35663
    },
    {
      "epoch": 14.24850179784259,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 35664
    },
    {
      "epoch": 14.248901318417898,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 35665
    },
    {
      "epoch": 14.249300838993209,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 35666
    },
    {
      "epoch": 14.249700359568518,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 35667
    },
    {
      "epoch": 14.250099880143827,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 35668
    },
    {
      "epoch": 14.250499400719137,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 35669
    },
    {
      "epoch": 14.250898921294446,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 35670
    },
    {
      "epoch": 14.251298441869757,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 35671
    },
    {
      "epoch": 14.251697962445066,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 35672
    },
    {
      "epoch": 14.252097483020375,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 35673
    },
    {
      "epoch": 14.252497003595685,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 35674
    },
    {
      "epoch": 14.252896524170994,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 35675
    },
    {
      "epoch": 14.253296044746305,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 35676
    },
    {
      "epoch": 14.253695565321614,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 35677
    },
    {
      "epoch": 14.254095085896923,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 35678
    },
    {
      "epoch": 14.254494606472234,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 35679
    },
    {
      "epoch": 14.254894127047542,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 35680
    },
    {
      "epoch": 14.255293647622853,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 35681
    },
    {
      "epoch": 14.255693168198162,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 35682
    },
    {
      "epoch": 14.256092688773473,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 35683
    },
    {
      "epoch": 14.256492209348782,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 35684
    },
    {
      "epoch": 14.25689172992409,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 35685
    },
    {
      "epoch": 14.257291250499401,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 35686
    },
    {
      "epoch": 14.25769077107471,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 35687
    },
    {
      "epoch": 14.25809029165002,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 35688
    },
    {
      "epoch": 14.25848981222533,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 35689
    },
    {
      "epoch": 14.258889332800639,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 35690
    },
    {
      "epoch": 14.25928885337595,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 35691
    },
    {
      "epoch": 14.259688373951258,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 35692
    },
    {
      "epoch": 14.260087894526569,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 35693
    },
    {
      "epoch": 14.260487415101878,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 35694
    },
    {
      "epoch": 14.260886935677187,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 35695
    },
    {
      "epoch": 14.261286456252497,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 35696
    },
    {
      "epoch": 14.261685976827806,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 35697
    },
    {
      "epoch": 14.262085497403117,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 35698
    },
    {
      "epoch": 14.262485017978426,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 35699
    },
    {
      "epoch": 14.262884538553735,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 35700
    },
    {
      "epoch": 14.263284059129045,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 35701
    },
    {
      "epoch": 14.263683579704354,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 35702
    },
    {
      "epoch": 14.264083100279665,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 35703
    },
    {
      "epoch": 14.264482620854974,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 35704
    },
    {
      "epoch": 14.264882141430284,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 35705
    },
    {
      "epoch": 14.265281662005593,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 35706
    },
    {
      "epoch": 14.265681182580902,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 35707
    },
    {
      "epoch": 14.266080703156213,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 35708
    },
    {
      "epoch": 14.266480223731522,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 35709
    },
    {
      "epoch": 14.266879744306832,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 35710
    },
    {
      "epoch": 14.267279264882141,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 35711
    },
    {
      "epoch": 14.26767878545745,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 35712
    },
    {
      "epoch": 14.268078306032761,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 35713
    },
    {
      "epoch": 14.26847782660807,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 35714
    },
    {
      "epoch": 14.26887734718338,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 35715
    },
    {
      "epoch": 14.26927686775869,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 35716
    },
    {
      "epoch": 14.269676388333998,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 35717
    },
    {
      "epoch": 14.270075908909309,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 35718
    },
    {
      "epoch": 14.270475429484618,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 35719
    },
    {
      "epoch": 14.270874950059929,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 35720
    },
    {
      "epoch": 14.271274470635237,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 35721
    },
    {
      "epoch": 14.271673991210548,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 35722
    },
    {
      "epoch": 14.272073511785857,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 35723
    },
    {
      "epoch": 14.272473032361166,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 35724
    },
    {
      "epoch": 14.272872552936477,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 35725
    },
    {
      "epoch": 14.273272073511786,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 35726
    },
    {
      "epoch": 14.273671594087096,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 35727
    },
    {
      "epoch": 14.274071114662405,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 35728
    },
    {
      "epoch": 14.274470635237714,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 35729
    },
    {
      "epoch": 14.274870155813025,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 35730
    },
    {
      "epoch": 14.275269676388334,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 35731
    },
    {
      "epoch": 14.275669196963644,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 35732
    },
    {
      "epoch": 14.276068717538953,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 35733
    },
    {
      "epoch": 14.276468238114262,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 35734
    },
    {
      "epoch": 14.276867758689573,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 35735
    },
    {
      "epoch": 14.277267279264882,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 35736
    },
    {
      "epoch": 14.277666799840192,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 35737
    },
    {
      "epoch": 14.278066320415501,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 35738
    },
    {
      "epoch": 14.278465840990812,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 35739
    },
    {
      "epoch": 14.27886536156612,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 35740
    },
    {
      "epoch": 14.27926488214143,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 35741
    },
    {
      "epoch": 14.27966440271674,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 35742
    },
    {
      "epoch": 14.28006392329205,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 35743
    },
    {
      "epoch": 14.28046344386736,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 35744
    },
    {
      "epoch": 14.280862964442669,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 35745
    },
    {
      "epoch": 14.281262485017978,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 35746
    },
    {
      "epoch": 14.281662005593288,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 35747
    },
    {
      "epoch": 14.282061526168597,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 35748
    },
    {
      "epoch": 14.282461046743908,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 35749
    },
    {
      "epoch": 14.282860567319217,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 35750
    },
    {
      "epoch": 14.283260087894526,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 35751
    },
    {
      "epoch": 14.283659608469836,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 35752
    },
    {
      "epoch": 14.284059129045145,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 35753
    },
    {
      "epoch": 14.284458649620456,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 35754
    },
    {
      "epoch": 14.284858170195765,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 35755
    },
    {
      "epoch": 14.285257690771076,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 35756
    },
    {
      "epoch": 14.285657211346384,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 35757
    },
    {
      "epoch": 14.286056731921693,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 35758
    },
    {
      "epoch": 14.286456252497004,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 35759
    },
    {
      "epoch": 14.286855773072313,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 35760
    },
    {
      "epoch": 14.287255293647624,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 35761
    },
    {
      "epoch": 14.287654814222932,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 35762
    },
    {
      "epoch": 14.288054334798241,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 35763
    },
    {
      "epoch": 14.288453855373552,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 35764
    },
    {
      "epoch": 14.288853375948861,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 35765
    },
    {
      "epoch": 14.289252896524172,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 35766
    },
    {
      "epoch": 14.28965241709948,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 35767
    },
    {
      "epoch": 14.29005193767479,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 35768
    },
    {
      "epoch": 14.2904514582501,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 35769
    },
    {
      "epoch": 14.290850978825409,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 35770
    },
    {
      "epoch": 14.29125049940072,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 35771
    },
    {
      "epoch": 14.291650019976029,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 35772
    },
    {
      "epoch": 14.29204954055134,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 35773
    },
    {
      "epoch": 14.292449061126648,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 35774
    },
    {
      "epoch": 14.292848581701957,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 35775
    },
    {
      "epoch": 14.293248102277268,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 35776
    },
    {
      "epoch": 14.293647622852577,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 35777
    },
    {
      "epoch": 14.294047143427887,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 35778
    },
    {
      "epoch": 14.294446664003196,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 35779
    },
    {
      "epoch": 14.294846184578505,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 35780
    },
    {
      "epoch": 14.295245705153816,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 35781
    },
    {
      "epoch": 14.295645225729125,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 35782
    },
    {
      "epoch": 14.296044746304435,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 35783
    },
    {
      "epoch": 14.296444266879744,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 35784
    },
    {
      "epoch": 14.296843787455053,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 35785
    },
    {
      "epoch": 14.297243308030364,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 35786
    },
    {
      "epoch": 14.297642828605673,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 35787
    },
    {
      "epoch": 14.298042349180983,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 35788
    },
    {
      "epoch": 14.298441869756292,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 35789
    },
    {
      "epoch": 14.298841390331603,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 35790
    },
    {
      "epoch": 14.299240910906912,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 35791
    },
    {
      "epoch": 14.29964043148222,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 35792
    },
    {
      "epoch": 14.300039952057531,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 35793
    },
    {
      "epoch": 14.30043947263284,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 35794
    },
    {
      "epoch": 14.300838993208151,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 35795
    },
    {
      "epoch": 14.30123851378346,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 35796
    },
    {
      "epoch": 14.301638034358769,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 35797
    },
    {
      "epoch": 14.30203755493408,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 35798
    },
    {
      "epoch": 14.302437075509388,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 35799
    },
    {
      "epoch": 14.302836596084699,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 35800
    },
    {
      "epoch": 14.303236116660008,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 35801
    },
    {
      "epoch": 14.303635637235317,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 35802
    },
    {
      "epoch": 14.304035157810628,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 35803
    },
    {
      "epoch": 14.304434678385936,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 35804
    },
    {
      "epoch": 14.304834198961247,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 35805
    },
    {
      "epoch": 14.305233719536556,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 35806
    },
    {
      "epoch": 14.305633240111867,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 35807
    },
    {
      "epoch": 14.306032760687176,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 35808
    },
    {
      "epoch": 14.306432281262484,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 35809
    },
    {
      "epoch": 14.306831801837795,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 35810
    },
    {
      "epoch": 14.307231322413104,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 35811
    },
    {
      "epoch": 14.307630842988415,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 35812
    },
    {
      "epoch": 14.308030363563724,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 35813
    },
    {
      "epoch": 14.308429884139032,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 35814
    },
    {
      "epoch": 14.308829404714343,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 35815
    },
    {
      "epoch": 14.309228925289652,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 35816
    },
    {
      "epoch": 14.309628445864963,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 35817
    },
    {
      "epoch": 14.310027966440272,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 35818
    },
    {
      "epoch": 14.31042748701558,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 35819
    },
    {
      "epoch": 14.310827007590891,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 35820
    },
    {
      "epoch": 14.3112265281662,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 35821
    },
    {
      "epoch": 14.31162604874151,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 35822
    },
    {
      "epoch": 14.31202556931682,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 35823
    },
    {
      "epoch": 14.31242508989213,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 35824
    },
    {
      "epoch": 14.31282461046744,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 35825
    },
    {
      "epoch": 14.313224131042748,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 35826
    },
    {
      "epoch": 14.313623651618059,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 35827
    },
    {
      "epoch": 14.314023172193368,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 35828
    },
    {
      "epoch": 14.314422692768678,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 35829
    },
    {
      "epoch": 14.314822213343987,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 35830
    },
    {
      "epoch": 14.315221733919296,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 35831
    },
    {
      "epoch": 14.315621254494607,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0099,
      "step": 35832
    },
    {
      "epoch": 14.316020775069916,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 35833
    },
    {
      "epoch": 14.316420295645226,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 35834
    },
    {
      "epoch": 14.316819816220535,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 35835
    },
    {
      "epoch": 14.317219336795844,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 35836
    },
    {
      "epoch": 14.317618857371155,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 35837
    },
    {
      "epoch": 14.318018377946464,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 35838
    },
    {
      "epoch": 14.318417898521774,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 35839
    },
    {
      "epoch": 14.318817419097083,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 35840
    },
    {
      "epoch": 14.319216939672392,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 35841
    },
    {
      "epoch": 14.319616460247703,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 35842
    },
    {
      "epoch": 14.320015980823012,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 35843
    },
    {
      "epoch": 14.320415501398323,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 35844
    },
    {
      "epoch": 14.320815021973631,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 35845
    },
    {
      "epoch": 14.321214542548942,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 35846
    },
    {
      "epoch": 14.321614063124251,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 35847
    },
    {
      "epoch": 14.32201358369956,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 35848
    },
    {
      "epoch": 14.32241310427487,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 35849
    },
    {
      "epoch": 14.32281262485018,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 35850
    },
    {
      "epoch": 14.32321214542549,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 35851
    },
    {
      "epoch": 14.323611666000799,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 35852
    },
    {
      "epoch": 14.324011186576108,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 35853
    },
    {
      "epoch": 14.324410707151419,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 35854
    },
    {
      "epoch": 14.324810227726728,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 35855
    },
    {
      "epoch": 14.325209748302038,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0084,
      "step": 35856
    },
    {
      "epoch": 14.325609268877347,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 35857
    },
    {
      "epoch": 14.326008789452656,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 35858
    },
    {
      "epoch": 14.326408310027967,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 35859
    },
    {
      "epoch": 14.326807830603276,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 35860
    },
    {
      "epoch": 14.327207351178586,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 35861
    },
    {
      "epoch": 14.327606871753895,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 35862
    },
    {
      "epoch": 14.328006392329206,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 35863
    },
    {
      "epoch": 14.328405912904515,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 35864
    },
    {
      "epoch": 14.328805433479824,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 35865
    },
    {
      "epoch": 14.329204954055134,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 35866
    },
    {
      "epoch": 14.329604474630443,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 35867
    },
    {
      "epoch": 14.330003995205754,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 35868
    },
    {
      "epoch": 14.330403515781063,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 35869
    },
    {
      "epoch": 14.330803036356372,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 35870
    },
    {
      "epoch": 14.331202556931682,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 35871
    },
    {
      "epoch": 14.331602077506991,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 35872
    },
    {
      "epoch": 14.332001598082302,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 35873
    },
    {
      "epoch": 14.33240111865761,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 35874
    },
    {
      "epoch": 14.33280063923292,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 35875
    },
    {
      "epoch": 14.33320015980823,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 35876
    },
    {
      "epoch": 14.33359968038354,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 35877
    },
    {
      "epoch": 14.33399920095885,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 35878
    },
    {
      "epoch": 14.334398721534159,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 35879
    },
    {
      "epoch": 14.334798242109468,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 35880
    },
    {
      "epoch": 14.335197762684778,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 35881
    },
    {
      "epoch": 14.335597283260087,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 35882
    },
    {
      "epoch": 14.335996803835398,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 35883
    },
    {
      "epoch": 14.336396324410707,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 35884
    },
    {
      "epoch": 14.336795844986018,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 35885
    },
    {
      "epoch": 14.337195365561326,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 35886
    },
    {
      "epoch": 14.337594886136635,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 35887
    },
    {
      "epoch": 14.337994406711946,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0467,
      "step": 35888
    },
    {
      "epoch": 14.338393927287255,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 35889
    },
    {
      "epoch": 14.338793447862566,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 35890
    },
    {
      "epoch": 14.339192968437874,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 35891
    },
    {
      "epoch": 14.339592489013183,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 35892
    },
    {
      "epoch": 14.339992009588494,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 35893
    },
    {
      "epoch": 14.340391530163803,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 35894
    },
    {
      "epoch": 14.340791050739114,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 35895
    },
    {
      "epoch": 14.341190571314423,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 35896
    },
    {
      "epoch": 14.341590091889731,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 35897
    },
    {
      "epoch": 14.341989612465042,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 35898
    },
    {
      "epoch": 14.342389133040351,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 35899
    },
    {
      "epoch": 14.342788653615662,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 35900
    },
    {
      "epoch": 14.34318817419097,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 35901
    },
    {
      "epoch": 14.343587694766281,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 35902
    },
    {
      "epoch": 14.34398721534159,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 35903
    },
    {
      "epoch": 14.344386735916899,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 35904
    },
    {
      "epoch": 14.34478625649221,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 35905
    },
    {
      "epoch": 14.345185777067519,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 35906
    },
    {
      "epoch": 14.34558529764283,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 35907
    },
    {
      "epoch": 14.345984818218138,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 35908
    },
    {
      "epoch": 14.346384338793447,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 35909
    },
    {
      "epoch": 14.346783859368758,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 35910
    },
    {
      "epoch": 14.347183379944067,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 35911
    },
    {
      "epoch": 14.347582900519377,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 35912
    },
    {
      "epoch": 14.347982421094686,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 35913
    },
    {
      "epoch": 14.348381941669995,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 35914
    },
    {
      "epoch": 14.348781462245306,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 35915
    },
    {
      "epoch": 14.349180982820615,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 35916
    },
    {
      "epoch": 14.349580503395925,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 35917
    },
    {
      "epoch": 14.349980023971234,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 35918
    },
    {
      "epoch": 14.350379544546545,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 35919
    },
    {
      "epoch": 14.350779065121854,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 35920
    },
    {
      "epoch": 14.351178585697163,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 35921
    },
    {
      "epoch": 14.351578106272473,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 35922
    },
    {
      "epoch": 14.351977626847782,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 35923
    },
    {
      "epoch": 14.352377147423093,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 35924
    },
    {
      "epoch": 14.352776667998402,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 35925
    },
    {
      "epoch": 14.35317618857371,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 35926
    },
    {
      "epoch": 14.353575709149021,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 35927
    },
    {
      "epoch": 14.35397522972433,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 35928
    },
    {
      "epoch": 14.354374750299641,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 35929
    },
    {
      "epoch": 14.35477427087495,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 35930
    },
    {
      "epoch": 14.355173791450259,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 35931
    },
    {
      "epoch": 14.35557331202557,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 35932
    },
    {
      "epoch": 14.355972832600878,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 35933
    },
    {
      "epoch": 14.356372353176189,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 35934
    },
    {
      "epoch": 14.356771873751498,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 35935
    },
    {
      "epoch": 14.357171394326809,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 35936
    },
    {
      "epoch": 14.357570914902118,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 35937
    },
    {
      "epoch": 14.357970435477426,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 35938
    },
    {
      "epoch": 14.358369956052737,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 35939
    },
    {
      "epoch": 14.358769476628046,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 35940
    },
    {
      "epoch": 14.359168997203357,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 35941
    },
    {
      "epoch": 14.359568517778666,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 35942
    },
    {
      "epoch": 14.359968038353975,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 35943
    },
    {
      "epoch": 14.360367558929285,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 35944
    },
    {
      "epoch": 14.360767079504594,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 35945
    },
    {
      "epoch": 14.361166600079905,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 35946
    },
    {
      "epoch": 14.361566120655214,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.007,
      "step": 35947
    },
    {
      "epoch": 14.361965641230523,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 35948
    },
    {
      "epoch": 14.362365161805833,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 35949
    },
    {
      "epoch": 14.362764682381142,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 35950
    },
    {
      "epoch": 14.363164202956453,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 35951
    },
    {
      "epoch": 14.363563723531762,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 35952
    },
    {
      "epoch": 14.363963244107072,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 35953
    },
    {
      "epoch": 14.364362764682381,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 35954
    },
    {
      "epoch": 14.36476228525769,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 35955
    },
    {
      "epoch": 14.365161805833,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 35956
    },
    {
      "epoch": 14.36556132640831,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 35957
    },
    {
      "epoch": 14.36596084698362,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 35958
    },
    {
      "epoch": 14.36636036755893,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 35959
    },
    {
      "epoch": 14.366759888134238,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 35960
    },
    {
      "epoch": 14.367159408709549,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 35961
    },
    {
      "epoch": 14.367558929284858,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 35962
    },
    {
      "epoch": 14.367958449860168,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 35963
    },
    {
      "epoch": 14.368357970435477,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 35964
    },
    {
      "epoch": 14.368757491010786,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 35965
    },
    {
      "epoch": 14.369157011586097,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 35966
    },
    {
      "epoch": 14.369556532161406,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 35967
    },
    {
      "epoch": 14.369956052736717,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 35968
    },
    {
      "epoch": 14.370355573312025,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 35969
    },
    {
      "epoch": 14.370755093887336,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 35970
    },
    {
      "epoch": 14.371154614462645,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 35971
    },
    {
      "epoch": 14.371554135037954,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 35972
    },
    {
      "epoch": 14.371953655613265,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 35973
    },
    {
      "epoch": 14.372353176188573,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 35974
    },
    {
      "epoch": 14.372752696763884,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 35975
    },
    {
      "epoch": 14.373152217339193,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 35976
    },
    {
      "epoch": 14.373551737914502,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 35977
    },
    {
      "epoch": 14.373951258489813,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 35978
    },
    {
      "epoch": 14.374350779065121,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 35979
    },
    {
      "epoch": 14.374750299640432,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 35980
    },
    {
      "epoch": 14.375149820215741,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 35981
    },
    {
      "epoch": 14.37554934079105,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 35982
    },
    {
      "epoch": 14.37594886136636,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 35983
    },
    {
      "epoch": 14.37634838194167,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 35984
    },
    {
      "epoch": 14.37674790251698,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 35985
    },
    {
      "epoch": 14.377147423092289,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 35986
    },
    {
      "epoch": 14.3775469436676,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 35987
    },
    {
      "epoch": 14.377946464242909,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 35988
    },
    {
      "epoch": 14.378345984818218,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 35989
    },
    {
      "epoch": 14.378745505393528,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 35990
    },
    {
      "epoch": 14.379145025968837,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 35991
    },
    {
      "epoch": 14.379544546544148,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 35992
    },
    {
      "epoch": 14.379944067119457,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 35993
    },
    {
      "epoch": 14.380343587694766,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 35994
    },
    {
      "epoch": 14.380743108270076,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 35995
    },
    {
      "epoch": 14.381142628845385,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 35996
    },
    {
      "epoch": 14.381542149420696,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 35997
    },
    {
      "epoch": 14.381941669996005,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 35998
    },
    {
      "epoch": 14.382341190571314,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 35999
    },
    {
      "epoch": 14.382740711146624,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 36000
    },
    {
      "epoch": 14.383140231721933,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 36001
    },
    {
      "epoch": 14.383539752297244,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 36002
    },
    {
      "epoch": 14.383939272872553,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 36003
    },
    {
      "epoch": 14.384338793447863,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 36004
    },
    {
      "epoch": 14.384738314023172,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 36005
    },
    {
      "epoch": 14.385137834598481,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 36006
    },
    {
      "epoch": 14.385537355173792,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 36007
    },
    {
      "epoch": 14.3859368757491,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 36008
    },
    {
      "epoch": 14.386336396324412,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 36009
    },
    {
      "epoch": 14.38673591689972,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 36010
    },
    {
      "epoch": 14.38713543747503,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 36011
    },
    {
      "epoch": 14.38753495805034,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 36012
    },
    {
      "epoch": 14.387934478625649,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 36013
    },
    {
      "epoch": 14.38833399920096,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 36014
    },
    {
      "epoch": 14.388733519776268,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 36015
    },
    {
      "epoch": 14.389133040351577,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 36016
    },
    {
      "epoch": 14.389532560926888,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 36017
    },
    {
      "epoch": 14.389932081502197,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 36018
    },
    {
      "epoch": 14.390331602077508,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 36019
    },
    {
      "epoch": 14.390731122652817,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 36020
    },
    {
      "epoch": 14.391130643228125,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 36021
    },
    {
      "epoch": 14.391530163803436,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 36022
    },
    {
      "epoch": 14.391929684378745,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 36023
    },
    {
      "epoch": 14.392329204954056,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 36024
    },
    {
      "epoch": 14.392728725529365,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 36025
    },
    {
      "epoch": 14.393128246104675,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 36026
    },
    {
      "epoch": 14.393527766679984,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 36027
    },
    {
      "epoch": 14.393927287255293,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 36028
    },
    {
      "epoch": 14.394326807830604,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 36029
    },
    {
      "epoch": 14.394726328405913,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 36030
    },
    {
      "epoch": 14.395125848981223,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 36031
    },
    {
      "epoch": 14.395525369556532,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 36032
    },
    {
      "epoch": 14.395924890131841,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 36033
    },
    {
      "epoch": 14.396324410707152,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 36034
    },
    {
      "epoch": 14.39672393128246,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 36035
    },
    {
      "epoch": 14.397123451857771,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 36036
    },
    {
      "epoch": 14.39752297243308,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 36037
    },
    {
      "epoch": 14.397922493008389,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 36038
    },
    {
      "epoch": 14.3983220135837,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 36039
    },
    {
      "epoch": 14.398721534159009,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 36040
    },
    {
      "epoch": 14.39912105473432,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 36041
    },
    {
      "epoch": 14.399520575309628,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 36042
    },
    {
      "epoch": 14.399920095884939,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 36043
    },
    {
      "epoch": 14.400319616460248,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 36044
    },
    {
      "epoch": 14.400719137035557,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 36045
    },
    {
      "epoch": 14.401118657610867,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 36046
    },
    {
      "epoch": 14.401518178186176,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 36047
    },
    {
      "epoch": 14.401917698761487,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 36048
    },
    {
      "epoch": 14.402317219336796,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 36049
    },
    {
      "epoch": 14.402716739912105,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 36050
    },
    {
      "epoch": 14.403116260487415,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 36051
    },
    {
      "epoch": 14.403515781062724,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 36052
    },
    {
      "epoch": 14.403915301638035,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 36053
    },
    {
      "epoch": 14.404314822213344,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 36054
    },
    {
      "epoch": 14.404714342788653,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 36055
    },
    {
      "epoch": 14.405113863363963,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 36056
    },
    {
      "epoch": 14.405513383939272,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 36057
    },
    {
      "epoch": 14.405912904514583,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 36058
    },
    {
      "epoch": 14.406312425089892,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 36059
    },
    {
      "epoch": 14.4067119456652,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 36060
    },
    {
      "epoch": 14.407111466240512,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 36061
    },
    {
      "epoch": 14.40751098681582,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 36062
    },
    {
      "epoch": 14.407910507391131,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 36063
    },
    {
      "epoch": 14.40831002796644,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 36064
    },
    {
      "epoch": 14.40870954854175,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 36065
    },
    {
      "epoch": 14.40910906911706,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 36066
    },
    {
      "epoch": 14.409508589692368,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 36067
    },
    {
      "epoch": 14.40990811026768,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 36068
    },
    {
      "epoch": 14.410307630842988,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 36069
    },
    {
      "epoch": 14.410707151418299,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 36070
    },
    {
      "epoch": 14.411106671993608,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 36071
    },
    {
      "epoch": 14.411506192568917,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 36072
    },
    {
      "epoch": 14.411905713144227,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 36073
    },
    {
      "epoch": 14.412305233719536,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 36074
    },
    {
      "epoch": 14.412704754294847,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 36075
    },
    {
      "epoch": 14.413104274870156,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 36076
    },
    {
      "epoch": 14.413503795445465,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 36077
    },
    {
      "epoch": 14.413903316020775,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 36078
    },
    {
      "epoch": 14.414302836596084,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 36079
    },
    {
      "epoch": 14.414702357171395,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 36080
    },
    {
      "epoch": 14.415101877746704,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 36081
    },
    {
      "epoch": 14.415501398322014,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 36082
    },
    {
      "epoch": 14.415900918897323,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 36083
    },
    {
      "epoch": 14.416300439472632,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 36084
    },
    {
      "epoch": 14.416699960047943,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 36085
    },
    {
      "epoch": 14.417099480623252,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 36086
    },
    {
      "epoch": 14.417499001198562,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 36087
    },
    {
      "epoch": 14.417898521773871,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 36088
    },
    {
      "epoch": 14.41829804234918,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 36089
    },
    {
      "epoch": 14.418697562924491,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 36090
    },
    {
      "epoch": 14.4190970834998,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 36091
    },
    {
      "epoch": 14.41949660407511,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 36092
    },
    {
      "epoch": 14.41989612465042,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 36093
    },
    {
      "epoch": 14.420295645225728,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 36094
    },
    {
      "epoch": 14.420695165801039,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 36095
    },
    {
      "epoch": 14.421094686376348,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 36096
    },
    {
      "epoch": 14.421494206951659,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 36097
    },
    {
      "epoch": 14.421893727526967,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 36098
    },
    {
      "epoch": 14.422293248102278,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 36099
    },
    {
      "epoch": 14.422692768677587,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 36100
    },
    {
      "epoch": 14.423092289252896,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 36101
    },
    {
      "epoch": 14.423491809828207,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 36102
    },
    {
      "epoch": 14.423891330403515,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 36103
    },
    {
      "epoch": 14.424290850978826,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 36104
    },
    {
      "epoch": 14.424690371554135,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 36105
    },
    {
      "epoch": 14.425089892129444,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 36106
    },
    {
      "epoch": 14.425489412704755,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 36107
    },
    {
      "epoch": 14.425888933280063,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 36108
    },
    {
      "epoch": 14.426288453855374,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 36109
    },
    {
      "epoch": 14.426687974430683,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 36110
    },
    {
      "epoch": 14.427087495005992,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 36111
    },
    {
      "epoch": 14.427487015581303,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 36112
    },
    {
      "epoch": 14.427886536156612,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 36113
    },
    {
      "epoch": 14.428286056731922,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 36114
    },
    {
      "epoch": 14.428685577307231,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 36115
    },
    {
      "epoch": 14.429085097882542,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 36116
    },
    {
      "epoch": 14.42948461845785,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 36117
    },
    {
      "epoch": 14.42988413903316,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 36118
    },
    {
      "epoch": 14.43028365960847,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 36119
    },
    {
      "epoch": 14.43068318018378,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 36120
    },
    {
      "epoch": 14.43108270075909,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 36121
    },
    {
      "epoch": 14.431482221334399,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 36122
    },
    {
      "epoch": 14.431881741909708,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 36123
    },
    {
      "epoch": 14.432281262485018,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 36124
    },
    {
      "epoch": 14.432680783060327,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 36125
    },
    {
      "epoch": 14.433080303635638,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 36126
    },
    {
      "epoch": 14.433479824210947,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 36127
    },
    {
      "epoch": 14.433879344786256,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 36128
    },
    {
      "epoch": 14.434278865361566,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 36129
    },
    {
      "epoch": 14.434678385936875,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 36130
    },
    {
      "epoch": 14.435077906512186,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 36131
    },
    {
      "epoch": 14.435477427087495,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 36132
    },
    {
      "epoch": 14.435876947662805,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 36133
    },
    {
      "epoch": 14.436276468238114,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 36134
    },
    {
      "epoch": 14.436675988813423,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 36135
    },
    {
      "epoch": 14.437075509388734,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 36136
    },
    {
      "epoch": 14.437475029964043,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 36137
    },
    {
      "epoch": 14.437874550539354,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 36138
    },
    {
      "epoch": 14.438274071114662,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 36139
    },
    {
      "epoch": 14.438673591689971,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 36140
    },
    {
      "epoch": 14.439073112265282,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 36141
    },
    {
      "epoch": 14.439472632840591,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 36142
    },
    {
      "epoch": 14.439872153415902,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 36143
    },
    {
      "epoch": 14.44027167399121,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 36144
    },
    {
      "epoch": 14.44067119456652,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 36145
    },
    {
      "epoch": 14.44107071514183,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 36146
    },
    {
      "epoch": 14.441470235717139,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 36147
    },
    {
      "epoch": 14.44186975629245,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 36148
    },
    {
      "epoch": 14.442269276867759,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 36149
    },
    {
      "epoch": 14.44266879744307,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 36150
    },
    {
      "epoch": 14.443068318018378,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 36151
    },
    {
      "epoch": 14.443467838593687,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 36152
    },
    {
      "epoch": 14.443867359168998,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 36153
    },
    {
      "epoch": 14.444266879744307,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 36154
    },
    {
      "epoch": 14.444666400319617,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 36155
    },
    {
      "epoch": 14.445065920894926,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 36156
    },
    {
      "epoch": 14.445465441470235,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 36157
    },
    {
      "epoch": 14.445864962045546,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 36158
    },
    {
      "epoch": 14.446264482620855,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 36159
    },
    {
      "epoch": 14.446664003196165,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 36160
    },
    {
      "epoch": 14.447063523771474,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 36161
    },
    {
      "epoch": 14.447463044346783,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0134,
      "step": 36162
    },
    {
      "epoch": 14.447862564922094,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 36163
    },
    {
      "epoch": 14.448262085497403,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 36164
    },
    {
      "epoch": 14.448661606072713,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 36165
    },
    {
      "epoch": 14.449061126648022,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 36166
    },
    {
      "epoch": 14.449460647223333,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 36167
    },
    {
      "epoch": 14.449860167798642,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 36168
    },
    {
      "epoch": 14.45025968837395,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 36169
    },
    {
      "epoch": 14.450659208949261,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 36170
    },
    {
      "epoch": 14.45105872952457,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 36171
    },
    {
      "epoch": 14.451458250099881,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 36172
    },
    {
      "epoch": 14.45185777067519,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 36173
    },
    {
      "epoch": 14.452257291250499,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 36174
    },
    {
      "epoch": 14.45265681182581,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 36175
    },
    {
      "epoch": 14.453056332401118,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 36176
    },
    {
      "epoch": 14.453455852976429,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 36177
    },
    {
      "epoch": 14.453855373551738,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 36178
    },
    {
      "epoch": 14.454254894127047,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 36179
    },
    {
      "epoch": 14.454654414702357,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 36180
    },
    {
      "epoch": 14.455053935277666,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 36181
    },
    {
      "epoch": 14.455453455852977,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 36182
    },
    {
      "epoch": 14.455852976428286,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 36183
    },
    {
      "epoch": 14.456252497003595,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 36184
    },
    {
      "epoch": 14.456652017578906,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 36185
    },
    {
      "epoch": 14.457051538154214,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 36186
    },
    {
      "epoch": 14.457451058729525,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 36187
    },
    {
      "epoch": 14.457850579304834,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 36188
    },
    {
      "epoch": 14.458250099880145,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 36189
    },
    {
      "epoch": 14.458649620455454,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 36190
    },
    {
      "epoch": 14.459049141030762,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 36191
    },
    {
      "epoch": 14.459448661606073,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 36192
    },
    {
      "epoch": 14.459848182181382,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 36193
    },
    {
      "epoch": 14.460247702756693,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 36194
    },
    {
      "epoch": 14.460647223332002,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 36195
    },
    {
      "epoch": 14.46104674390731,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 36196
    },
    {
      "epoch": 14.461446264482621,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 36197
    },
    {
      "epoch": 14.46184578505793,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 36198
    },
    {
      "epoch": 14.46224530563324,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 36199
    },
    {
      "epoch": 14.46264482620855,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 36200
    },
    {
      "epoch": 14.463044346783859,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 36201
    },
    {
      "epoch": 14.46344386735917,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 36202
    },
    {
      "epoch": 14.463843387934478,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 36203
    },
    {
      "epoch": 14.464242908509789,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 36204
    },
    {
      "epoch": 14.464642429085098,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 36205
    },
    {
      "epoch": 14.465041949660408,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 36206
    },
    {
      "epoch": 14.465441470235717,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 36207
    },
    {
      "epoch": 14.465840990811026,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 36208
    },
    {
      "epoch": 14.466240511386337,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 36209
    },
    {
      "epoch": 14.466640031961646,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 36210
    },
    {
      "epoch": 14.467039552536956,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 36211
    },
    {
      "epoch": 14.467439073112265,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 36212
    },
    {
      "epoch": 14.467838593687574,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 36213
    },
    {
      "epoch": 14.468238114262885,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 36214
    },
    {
      "epoch": 14.468637634838194,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 36215
    },
    {
      "epoch": 14.469037155413504,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 36216
    },
    {
      "epoch": 14.469436675988813,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 36217
    },
    {
      "epoch": 14.469836196564122,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 36218
    },
    {
      "epoch": 14.470235717139433,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 36219
    },
    {
      "epoch": 14.470635237714742,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 36220
    },
    {
      "epoch": 14.471034758290052,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 36221
    },
    {
      "epoch": 14.471434278865361,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 36222
    },
    {
      "epoch": 14.471833799440672,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 36223
    },
    {
      "epoch": 14.472233320015981,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 36224
    },
    {
      "epoch": 14.47263284059129,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 36225
    },
    {
      "epoch": 14.4730323611666,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 36226
    },
    {
      "epoch": 14.47343188174191,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 36227
    },
    {
      "epoch": 14.47383140231722,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 36228
    },
    {
      "epoch": 14.474230922892529,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 36229
    },
    {
      "epoch": 14.474630443467838,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 36230
    },
    {
      "epoch": 14.475029964043149,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 36231
    },
    {
      "epoch": 14.475429484618457,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 36232
    },
    {
      "epoch": 14.475829005193768,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 36233
    },
    {
      "epoch": 14.476228525769077,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 36234
    },
    {
      "epoch": 14.476628046344386,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 36235
    },
    {
      "epoch": 14.477027566919697,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 36236
    },
    {
      "epoch": 14.477427087495006,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 36237
    },
    {
      "epoch": 14.477826608070316,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 36238
    },
    {
      "epoch": 14.478226128645625,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 36239
    },
    {
      "epoch": 14.478625649220934,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 36240
    },
    {
      "epoch": 14.479025169796245,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 36241
    },
    {
      "epoch": 14.479424690371554,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 36242
    },
    {
      "epoch": 14.479824210946864,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 36243
    },
    {
      "epoch": 14.480223731522173,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 36244
    },
    {
      "epoch": 14.480623252097484,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 36245
    },
    {
      "epoch": 14.481022772672793,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 36246
    },
    {
      "epoch": 14.481422293248102,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 36247
    },
    {
      "epoch": 14.481821813823412,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 36248
    },
    {
      "epoch": 14.482221334398721,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 36249
    },
    {
      "epoch": 14.482620854974032,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 36250
    },
    {
      "epoch": 14.48302037554934,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 36251
    },
    {
      "epoch": 14.48341989612465,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 36252
    },
    {
      "epoch": 14.48381941669996,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 36253
    },
    {
      "epoch": 14.48421893727527,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 36254
    },
    {
      "epoch": 14.48461845785058,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 36255
    },
    {
      "epoch": 14.485017978425889,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 36256
    },
    {
      "epoch": 14.485417499001198,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 36257
    },
    {
      "epoch": 14.485817019576508,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 36258
    },
    {
      "epoch": 14.486216540151817,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 36259
    },
    {
      "epoch": 14.486616060727128,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 36260
    },
    {
      "epoch": 14.487015581302437,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 36261
    },
    {
      "epoch": 14.487415101877748,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 36262
    },
    {
      "epoch": 14.487814622453056,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 36263
    },
    {
      "epoch": 14.488214143028365,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 36264
    },
    {
      "epoch": 14.488613663603676,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 36265
    },
    {
      "epoch": 14.489013184178985,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 36266
    },
    {
      "epoch": 14.489412704754296,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 36267
    },
    {
      "epoch": 14.489812225329604,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 36268
    },
    {
      "epoch": 14.490211745904913,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 36269
    },
    {
      "epoch": 14.490611266480224,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 36270
    },
    {
      "epoch": 14.491010787055533,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 36271
    },
    {
      "epoch": 14.491410307630844,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 36272
    },
    {
      "epoch": 14.491809828206152,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 36273
    },
    {
      "epoch": 14.492209348781461,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 36274
    },
    {
      "epoch": 14.492608869356772,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 36275
    },
    {
      "epoch": 14.493008389932081,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 36276
    },
    {
      "epoch": 14.493407910507392,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 36277
    },
    {
      "epoch": 14.4938074310827,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 36278
    },
    {
      "epoch": 14.494206951658011,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 36279
    },
    {
      "epoch": 14.49460647223332,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 36280
    },
    {
      "epoch": 14.495005992808629,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 36281
    },
    {
      "epoch": 14.49540551338394,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 36282
    },
    {
      "epoch": 14.495805033959249,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 36283
    },
    {
      "epoch": 14.49620455453456,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 36284
    },
    {
      "epoch": 14.496604075109868,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 36285
    },
    {
      "epoch": 14.497003595685177,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 36286
    },
    {
      "epoch": 14.497403116260488,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 36287
    },
    {
      "epoch": 14.497802636835797,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 36288
    },
    {
      "epoch": 14.498202157411107,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 36289
    },
    {
      "epoch": 14.498601677986416,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 36290
    },
    {
      "epoch": 14.499001198561725,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 36291
    },
    {
      "epoch": 14.499400719137036,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 36292
    },
    {
      "epoch": 14.499800239712345,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 36293
    },
    {
      "epoch": 14.500199760287655,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 36294
    },
    {
      "epoch": 14.500599280862964,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 36295
    },
    {
      "epoch": 14.500998801438275,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 36296
    },
    {
      "epoch": 14.501398322013584,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 36297
    },
    {
      "epoch": 14.501797842588893,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 36298
    },
    {
      "epoch": 14.502197363164203,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 36299
    },
    {
      "epoch": 14.502596883739512,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 36300
    },
    {
      "epoch": 14.502996404314823,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 36301
    },
    {
      "epoch": 14.503395924890132,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 36302
    },
    {
      "epoch": 14.50379544546544,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 36303
    },
    {
      "epoch": 14.504194966040751,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 36304
    },
    {
      "epoch": 14.50459448661606,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 36305
    },
    {
      "epoch": 14.504994007191371,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 36306
    },
    {
      "epoch": 14.50539352776668,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 36307
    },
    {
      "epoch": 14.505793048341989,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 36308
    },
    {
      "epoch": 14.5061925689173,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 36309
    },
    {
      "epoch": 14.506592089492608,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 36310
    },
    {
      "epoch": 14.506991610067919,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 36311
    },
    {
      "epoch": 14.507391130643228,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 36312
    },
    {
      "epoch": 14.507790651218539,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 36313
    },
    {
      "epoch": 14.508190171793848,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 36314
    },
    {
      "epoch": 14.508589692369156,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 36315
    },
    {
      "epoch": 14.508989212944467,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 36316
    },
    {
      "epoch": 14.509388733519776,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 36317
    },
    {
      "epoch": 14.509788254095087,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 36318
    },
    {
      "epoch": 14.510187774670396,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 36319
    },
    {
      "epoch": 14.510587295245704,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 36320
    },
    {
      "epoch": 14.510986815821015,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 36321
    },
    {
      "epoch": 14.511386336396324,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 36322
    },
    {
      "epoch": 14.511785856971635,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 36323
    },
    {
      "epoch": 14.512185377546944,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 36324
    },
    {
      "epoch": 14.512584898122252,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 36325
    },
    {
      "epoch": 14.512984418697563,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 36326
    },
    {
      "epoch": 14.513383939272872,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 36327
    },
    {
      "epoch": 14.513783459848183,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 36328
    },
    {
      "epoch": 14.514182980423492,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 36329
    },
    {
      "epoch": 14.514582500998802,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 36330
    },
    {
      "epoch": 14.514982021574111,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 36331
    },
    {
      "epoch": 14.51538154214942,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 36332
    },
    {
      "epoch": 14.51578106272473,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 36333
    },
    {
      "epoch": 14.51618058330004,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 36334
    },
    {
      "epoch": 14.51658010387535,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 36335
    },
    {
      "epoch": 14.51697962445066,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 36336
    },
    {
      "epoch": 14.517379145025968,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 36337
    },
    {
      "epoch": 14.517778665601279,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 36338
    },
    {
      "epoch": 14.518178186176588,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 36339
    },
    {
      "epoch": 14.518577706751898,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 36340
    },
    {
      "epoch": 14.518977227327207,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 36341
    },
    {
      "epoch": 14.519376747902516,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 36342
    },
    {
      "epoch": 14.519776268477827,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 36343
    },
    {
      "epoch": 14.520175789053136,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 36344
    },
    {
      "epoch": 14.520575309628446,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 36345
    },
    {
      "epoch": 14.520974830203755,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 36346
    },
    {
      "epoch": 14.521374350779066,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 36347
    },
    {
      "epoch": 14.521773871354375,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 36348
    },
    {
      "epoch": 14.522173391929684,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 36349
    },
    {
      "epoch": 14.522572912504994,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 36350
    },
    {
      "epoch": 14.522972433080303,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 36351
    },
    {
      "epoch": 14.523371953655614,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 36352
    },
    {
      "epoch": 14.523771474230923,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 36353
    },
    {
      "epoch": 14.524170994806232,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 36354
    },
    {
      "epoch": 14.524570515381543,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 36355
    },
    {
      "epoch": 14.524970035956851,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 36356
    },
    {
      "epoch": 14.525369556532162,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 36357
    },
    {
      "epoch": 14.525769077107471,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 36358
    },
    {
      "epoch": 14.52616859768278,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 36359
    },
    {
      "epoch": 14.52656811825809,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 36360
    },
    {
      "epoch": 14.5269676388334,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 36361
    },
    {
      "epoch": 14.52736715940871,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 36362
    },
    {
      "epoch": 14.527766679984019,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 36363
    },
    {
      "epoch": 14.52816620055933,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 36364
    },
    {
      "epoch": 14.528565721134639,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 36365
    },
    {
      "epoch": 14.528965241709948,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 36366
    },
    {
      "epoch": 14.529364762285258,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 36367
    },
    {
      "epoch": 14.529764282860567,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 36368
    },
    {
      "epoch": 14.530163803435878,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 36369
    },
    {
      "epoch": 14.530563324011187,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 36370
    },
    {
      "epoch": 14.530962844586496,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 36371
    },
    {
      "epoch": 14.531362365161806,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 36372
    },
    {
      "epoch": 14.531761885737115,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 36373
    },
    {
      "epoch": 14.532161406312426,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 36374
    },
    {
      "epoch": 14.532560926887735,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 36375
    },
    {
      "epoch": 14.532960447463044,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 36376
    },
    {
      "epoch": 14.533359968038354,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 36377
    },
    {
      "epoch": 14.533759488613663,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 36378
    },
    {
      "epoch": 14.534159009188974,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 36379
    },
    {
      "epoch": 14.534558529764283,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 36380
    },
    {
      "epoch": 14.534958050339593,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 36381
    },
    {
      "epoch": 14.535357570914902,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 36382
    },
    {
      "epoch": 14.535757091490211,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 36383
    },
    {
      "epoch": 14.536156612065522,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 36384
    },
    {
      "epoch": 14.53655613264083,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 36385
    },
    {
      "epoch": 14.53695565321614,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 36386
    },
    {
      "epoch": 14.53735517379145,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 36387
    },
    {
      "epoch": 14.53775469436676,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 36388
    },
    {
      "epoch": 14.53815421494207,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 36389
    },
    {
      "epoch": 14.538553735517379,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 36390
    },
    {
      "epoch": 14.53895325609269,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 36391
    },
    {
      "epoch": 14.539352776667998,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 36392
    },
    {
      "epoch": 14.539752297243307,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 36393
    },
    {
      "epoch": 14.540151817818618,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 36394
    },
    {
      "epoch": 14.540551338393927,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 36395
    },
    {
      "epoch": 14.540950858969238,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 36396
    },
    {
      "epoch": 14.541350379544546,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 36397
    },
    {
      "epoch": 14.541749900119855,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 36398
    },
    {
      "epoch": 14.542149420695166,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 36399
    },
    {
      "epoch": 14.542548941270475,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 36400
    },
    {
      "epoch": 14.542948461845786,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 36401
    },
    {
      "epoch": 14.543347982421094,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 36402
    },
    {
      "epoch": 14.543747502996403,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 36403
    },
    {
      "epoch": 14.544147023571714,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 36404
    },
    {
      "epoch": 14.544546544147023,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 36405
    },
    {
      "epoch": 14.544946064722334,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 36406
    },
    {
      "epoch": 14.545345585297643,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 36407
    },
    {
      "epoch": 14.545745105872953,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 36408
    },
    {
      "epoch": 14.546144626448262,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 36409
    },
    {
      "epoch": 14.546544147023571,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 36410
    },
    {
      "epoch": 14.546943667598882,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 36411
    },
    {
      "epoch": 14.54734318817419,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 36412
    },
    {
      "epoch": 14.547742708749501,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 36413
    },
    {
      "epoch": 14.54814222932481,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 36414
    },
    {
      "epoch": 14.548541749900119,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 36415
    },
    {
      "epoch": 14.54894127047543,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 36416
    },
    {
      "epoch": 14.549340791050739,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 36417
    },
    {
      "epoch": 14.54974031162605,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 36418
    },
    {
      "epoch": 14.550139832201358,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 36419
    },
    {
      "epoch": 14.550539352776667,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 36420
    },
    {
      "epoch": 14.550938873351978,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 36421
    },
    {
      "epoch": 14.551338393927287,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 36422
    },
    {
      "epoch": 14.551737914502597,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0094,
      "step": 36423
    },
    {
      "epoch": 14.552137435077906,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 36424
    },
    {
      "epoch": 14.552536955653217,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 36425
    },
    {
      "epoch": 14.552936476228526,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 36426
    },
    {
      "epoch": 14.553335996803835,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 36427
    },
    {
      "epoch": 14.553735517379145,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 36428
    },
    {
      "epoch": 14.554135037954454,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 36429
    },
    {
      "epoch": 14.554534558529765,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 36430
    },
    {
      "epoch": 14.554934079105074,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 36431
    },
    {
      "epoch": 14.555333599680383,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 36432
    },
    {
      "epoch": 14.555733120255693,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 36433
    },
    {
      "epoch": 14.556132640831002,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 36434
    },
    {
      "epoch": 14.556532161406313,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 36435
    },
    {
      "epoch": 14.556931681981622,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 36436
    },
    {
      "epoch": 14.55733120255693,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 36437
    },
    {
      "epoch": 14.557730723132241,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 36438
    },
    {
      "epoch": 14.55813024370755,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 36439
    },
    {
      "epoch": 14.558529764282861,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 36440
    },
    {
      "epoch": 14.55892928485817,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 36441
    },
    {
      "epoch": 14.55932880543348,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 36442
    },
    {
      "epoch": 14.55972832600879,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 36443
    },
    {
      "epoch": 14.560127846584098,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 36444
    },
    {
      "epoch": 14.560527367159409,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 36445
    },
    {
      "epoch": 14.560926887734718,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 36446
    },
    {
      "epoch": 14.561326408310029,
      "grad_norm": 0.036376953125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 36447
    },
    {
      "epoch": 14.561725928885338,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 36448
    },
    {
      "epoch": 14.562125449460646,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 36449
    },
    {
      "epoch": 14.562524970035957,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 36450
    },
    {
      "epoch": 14.562924490611266,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 36451
    },
    {
      "epoch": 14.563324011186577,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 36452
    },
    {
      "epoch": 14.563723531761886,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 36453
    },
    {
      "epoch": 14.564123052337195,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 36454
    },
    {
      "epoch": 14.564522572912505,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 36455
    },
    {
      "epoch": 14.564922093487814,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 36456
    },
    {
      "epoch": 14.565321614063125,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 36457
    },
    {
      "epoch": 14.565721134638434,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 36458
    },
    {
      "epoch": 14.566120655213744,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 36459
    },
    {
      "epoch": 14.566520175789053,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 36460
    },
    {
      "epoch": 14.566919696364362,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 36461
    },
    {
      "epoch": 14.567319216939673,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 36462
    },
    {
      "epoch": 14.567718737514982,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 36463
    },
    {
      "epoch": 14.568118258090292,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 36464
    },
    {
      "epoch": 14.568517778665601,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 36465
    },
    {
      "epoch": 14.56891729924091,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 36466
    },
    {
      "epoch": 14.56931681981622,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 36467
    },
    {
      "epoch": 14.56971634039153,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 36468
    },
    {
      "epoch": 14.57011586096684,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 36469
    },
    {
      "epoch": 14.57051538154215,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 36470
    },
    {
      "epoch": 14.570914902117458,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 36471
    },
    {
      "epoch": 14.571314422692769,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 36472
    },
    {
      "epoch": 14.571713943268078,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 36473
    },
    {
      "epoch": 14.572113463843388,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 36474
    },
    {
      "epoch": 14.572512984418697,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 36475
    },
    {
      "epoch": 14.572912504994008,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 36476
    },
    {
      "epoch": 14.573312025569317,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 36477
    },
    {
      "epoch": 14.573711546144626,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 36478
    },
    {
      "epoch": 14.574111066719937,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 36479
    },
    {
      "epoch": 14.574510587295245,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 36480
    },
    {
      "epoch": 14.574910107870556,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 36481
    },
    {
      "epoch": 14.575309628445865,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 36482
    },
    {
      "epoch": 14.575709149021174,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 36483
    },
    {
      "epoch": 14.576108669596485,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 36484
    },
    {
      "epoch": 14.576508190171793,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 36485
    },
    {
      "epoch": 14.576907710747104,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 36486
    },
    {
      "epoch": 14.577307231322413,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 36487
    },
    {
      "epoch": 14.577706751897722,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 36488
    },
    {
      "epoch": 14.578106272473033,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 36489
    },
    {
      "epoch": 14.578505793048341,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 36490
    },
    {
      "epoch": 14.578905313623652,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 36491
    },
    {
      "epoch": 14.579304834198961,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 36492
    },
    {
      "epoch": 14.579704354774272,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 36493
    },
    {
      "epoch": 14.58010387534958,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 36494
    },
    {
      "epoch": 14.58050339592489,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 36495
    },
    {
      "epoch": 14.5809029165002,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 36496
    },
    {
      "epoch": 14.581302437075509,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 36497
    },
    {
      "epoch": 14.58170195765082,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 36498
    },
    {
      "epoch": 14.582101478226129,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 36499
    },
    {
      "epoch": 14.582500998801438,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 36500
    },
    {
      "epoch": 14.582900519376748,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 36501
    },
    {
      "epoch": 14.583300039952057,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 36502
    },
    {
      "epoch": 14.583699560527368,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 36503
    },
    {
      "epoch": 14.584099081102677,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 36504
    },
    {
      "epoch": 14.584498601677986,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 36505
    },
    {
      "epoch": 14.584898122253296,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 36506
    },
    {
      "epoch": 14.585297642828605,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 36507
    },
    {
      "epoch": 14.585697163403916,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 36508
    },
    {
      "epoch": 14.586096683979225,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 36509
    },
    {
      "epoch": 14.586496204554535,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 36510
    },
    {
      "epoch": 14.586895725129844,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 36511
    },
    {
      "epoch": 14.587295245705153,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0108,
      "step": 36512
    },
    {
      "epoch": 14.587694766280464,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 36513
    },
    {
      "epoch": 14.588094286855773,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 36514
    },
    {
      "epoch": 14.588493807431083,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 36515
    },
    {
      "epoch": 14.588893328006392,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 36516
    },
    {
      "epoch": 14.589292848581701,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 36517
    },
    {
      "epoch": 14.589692369157012,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 36518
    },
    {
      "epoch": 14.59009188973232,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 36519
    },
    {
      "epoch": 14.590491410307632,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 36520
    },
    {
      "epoch": 14.59089093088294,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 36521
    },
    {
      "epoch": 14.59129045145825,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 36522
    },
    {
      "epoch": 14.59168997203356,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 36523
    },
    {
      "epoch": 14.592089492608869,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 36524
    },
    {
      "epoch": 14.59248901318418,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 36525
    },
    {
      "epoch": 14.592888533759488,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 36526
    },
    {
      "epoch": 14.5932880543348,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 36527
    },
    {
      "epoch": 14.593687574910108,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 36528
    },
    {
      "epoch": 14.594087095485417,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 36529
    },
    {
      "epoch": 14.594486616060728,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 36530
    },
    {
      "epoch": 14.594886136636037,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 36531
    },
    {
      "epoch": 14.595285657211347,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 36532
    },
    {
      "epoch": 14.595685177786656,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 36533
    },
    {
      "epoch": 14.596084698361965,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 36534
    },
    {
      "epoch": 14.596484218937276,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 36535
    },
    {
      "epoch": 14.596883739512585,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 36536
    },
    {
      "epoch": 14.597283260087895,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 36537
    },
    {
      "epoch": 14.597682780663204,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 36538
    },
    {
      "epoch": 14.598082301238513,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 36539
    },
    {
      "epoch": 14.598481821813824,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 36540
    },
    {
      "epoch": 14.598881342389133,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 36541
    },
    {
      "epoch": 14.599280862964443,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 36542
    },
    {
      "epoch": 14.599680383539752,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 36543
    },
    {
      "epoch": 14.600079904115063,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 36544
    },
    {
      "epoch": 14.600479424690372,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 36545
    },
    {
      "epoch": 14.60087894526568,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 36546
    },
    {
      "epoch": 14.601278465840991,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 36547
    },
    {
      "epoch": 14.6016779864163,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 36548
    },
    {
      "epoch": 14.602077506991611,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 36549
    },
    {
      "epoch": 14.60247702756692,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 36550
    },
    {
      "epoch": 14.602876548142229,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 36551
    },
    {
      "epoch": 14.60327606871754,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 36552
    },
    {
      "epoch": 14.603675589292848,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 36553
    },
    {
      "epoch": 14.604075109868159,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 36554
    },
    {
      "epoch": 14.604474630443468,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 36555
    },
    {
      "epoch": 14.604874151018777,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 36556
    },
    {
      "epoch": 14.605273671594087,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 36557
    },
    {
      "epoch": 14.605673192169396,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 36558
    },
    {
      "epoch": 14.606072712744707,
      "grad_norm": 0.171875,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 36559
    },
    {
      "epoch": 14.606472233320016,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 36560
    },
    {
      "epoch": 14.606871753895327,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 36561
    },
    {
      "epoch": 14.607271274470635,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 36562
    },
    {
      "epoch": 14.607670795045944,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 36563
    },
    {
      "epoch": 14.608070315621255,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 36564
    },
    {
      "epoch": 14.608469836196564,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 36565
    },
    {
      "epoch": 14.608869356771873,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 36566
    },
    {
      "epoch": 14.609268877347183,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 36567
    },
    {
      "epoch": 14.609668397922492,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 36568
    },
    {
      "epoch": 14.610067918497803,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 36569
    },
    {
      "epoch": 14.610467439073112,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 36570
    },
    {
      "epoch": 14.610866959648423,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 36571
    },
    {
      "epoch": 14.611266480223732,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 36572
    },
    {
      "epoch": 14.61166600079904,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 36573
    },
    {
      "epoch": 14.612065521374351,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 36574
    },
    {
      "epoch": 14.61246504194966,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 36575
    },
    {
      "epoch": 14.61286456252497,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 36576
    },
    {
      "epoch": 14.61326408310028,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 36577
    },
    {
      "epoch": 14.613663603675588,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 36578
    },
    {
      "epoch": 14.6140631242509,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 36579
    },
    {
      "epoch": 14.614462644826208,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 36580
    },
    {
      "epoch": 14.614862165401519,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 36581
    },
    {
      "epoch": 14.615261685976828,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 36582
    },
    {
      "epoch": 14.615661206552137,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 36583
    },
    {
      "epoch": 14.616060727127447,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 36584
    },
    {
      "epoch": 14.616460247702756,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 36585
    },
    {
      "epoch": 14.616859768278067,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 36586
    },
    {
      "epoch": 14.617259288853376,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0128,
      "step": 36587
    },
    {
      "epoch": 14.617658809428686,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 36588
    },
    {
      "epoch": 14.618058330003995,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 36589
    },
    {
      "epoch": 14.618457850579304,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 36590
    },
    {
      "epoch": 14.618857371154615,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 36591
    },
    {
      "epoch": 14.619256891729924,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 36592
    },
    {
      "epoch": 14.619656412305234,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 36593
    },
    {
      "epoch": 14.620055932880543,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 36594
    },
    {
      "epoch": 14.620455453455852,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 36595
    },
    {
      "epoch": 14.620854974031163,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 36596
    },
    {
      "epoch": 14.621254494606472,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 36597
    },
    {
      "epoch": 14.621654015181782,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 36598
    },
    {
      "epoch": 14.622053535757091,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 36599
    },
    {
      "epoch": 14.6224530563324,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 36600
    },
    {
      "epoch": 14.622852576907711,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 36601
    },
    {
      "epoch": 14.62325209748302,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 36602
    },
    {
      "epoch": 14.62365161805833,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 36603
    },
    {
      "epoch": 14.62405113863364,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 36604
    },
    {
      "epoch": 14.62445065920895,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 36605
    },
    {
      "epoch": 14.624850179784259,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 36606
    },
    {
      "epoch": 14.625249700359568,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 36607
    },
    {
      "epoch": 14.625649220934879,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 36608
    },
    {
      "epoch": 14.626048741510187,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 36609
    },
    {
      "epoch": 14.626448262085498,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 36610
    },
    {
      "epoch": 14.626847782660807,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 36611
    },
    {
      "epoch": 14.627247303236116,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 36612
    },
    {
      "epoch": 14.627646823811427,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 36613
    },
    {
      "epoch": 14.628046344386735,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 36614
    },
    {
      "epoch": 14.628445864962046,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 36615
    },
    {
      "epoch": 14.628845385537355,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 36616
    },
    {
      "epoch": 14.629244906112664,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 36617
    },
    {
      "epoch": 14.629644426687975,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 36618
    },
    {
      "epoch": 14.630043947263283,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 36619
    },
    {
      "epoch": 14.630443467838594,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 36620
    },
    {
      "epoch": 14.630842988413903,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 36621
    },
    {
      "epoch": 14.631242508989214,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 36622
    },
    {
      "epoch": 14.631642029564523,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 36623
    },
    {
      "epoch": 14.632041550139832,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 36624
    },
    {
      "epoch": 14.632441070715142,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 36625
    },
    {
      "epoch": 14.632840591290451,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 36626
    },
    {
      "epoch": 14.633240111865762,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 36627
    },
    {
      "epoch": 14.63363963244107,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 36628
    },
    {
      "epoch": 14.63403915301638,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 36629
    },
    {
      "epoch": 14.63443867359169,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 36630
    },
    {
      "epoch": 14.634838194167,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 36631
    },
    {
      "epoch": 14.63523771474231,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 36632
    },
    {
      "epoch": 14.635637235317619,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 36633
    },
    {
      "epoch": 14.636036755892928,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 36634
    },
    {
      "epoch": 14.636436276468238,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 36635
    },
    {
      "epoch": 14.636835797043547,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 36636
    },
    {
      "epoch": 14.637235317618858,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 36637
    },
    {
      "epoch": 14.637634838194167,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 36638
    },
    {
      "epoch": 14.638034358769477,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 36639
    },
    {
      "epoch": 14.638433879344786,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 36640
    },
    {
      "epoch": 14.638833399920095,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 36641
    },
    {
      "epoch": 14.639232920495406,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 36642
    },
    {
      "epoch": 14.639632441070715,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 36643
    },
    {
      "epoch": 14.640031961646025,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 36644
    },
    {
      "epoch": 14.640431482221334,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 36645
    },
    {
      "epoch": 14.640831002796643,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 36646
    },
    {
      "epoch": 14.641230523371954,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 36647
    },
    {
      "epoch": 14.641630043947263,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 36648
    },
    {
      "epoch": 14.642029564522574,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 36649
    },
    {
      "epoch": 14.642429085097882,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 36650
    },
    {
      "epoch": 14.642828605673191,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 36651
    },
    {
      "epoch": 14.643228126248502,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 36652
    },
    {
      "epoch": 14.643627646823811,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 36653
    },
    {
      "epoch": 14.644027167399122,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 36654
    },
    {
      "epoch": 14.64442668797443,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 36655
    },
    {
      "epoch": 14.644826208549741,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 36656
    },
    {
      "epoch": 14.64522572912505,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 36657
    },
    {
      "epoch": 14.645625249700359,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 36658
    },
    {
      "epoch": 14.64602477027567,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 36659
    },
    {
      "epoch": 14.646424290850979,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 36660
    },
    {
      "epoch": 14.64682381142629,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 36661
    },
    {
      "epoch": 14.647223332001598,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 36662
    },
    {
      "epoch": 14.647622852576907,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 36663
    },
    {
      "epoch": 14.648022373152218,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 36664
    },
    {
      "epoch": 14.648421893727527,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 36665
    },
    {
      "epoch": 14.648821414302837,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 36666
    },
    {
      "epoch": 14.649220934878146,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 36667
    },
    {
      "epoch": 14.649620455453455,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 36668
    },
    {
      "epoch": 14.650019976028766,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 36669
    },
    {
      "epoch": 14.650419496604075,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 36670
    },
    {
      "epoch": 14.650819017179385,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 36671
    },
    {
      "epoch": 14.651218537754694,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 36672
    },
    {
      "epoch": 14.651618058330005,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 36673
    },
    {
      "epoch": 14.652017578905314,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 36674
    },
    {
      "epoch": 14.652417099480623,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 36675
    },
    {
      "epoch": 14.652816620055933,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 36676
    },
    {
      "epoch": 14.653216140631242,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 36677
    },
    {
      "epoch": 14.653615661206553,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 36678
    },
    {
      "epoch": 14.654015181781862,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 36679
    },
    {
      "epoch": 14.65441470235717,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 36680
    },
    {
      "epoch": 14.654814222932481,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 36681
    },
    {
      "epoch": 14.65521374350779,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 36682
    },
    {
      "epoch": 14.655613264083101,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 36683
    },
    {
      "epoch": 14.65601278465841,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 36684
    },
    {
      "epoch": 14.656412305233719,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 36685
    },
    {
      "epoch": 14.65681182580903,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 36686
    },
    {
      "epoch": 14.657211346384338,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 36687
    },
    {
      "epoch": 14.657610866959649,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 36688
    },
    {
      "epoch": 14.658010387534958,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 36689
    },
    {
      "epoch": 14.658409908110269,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 36690
    },
    {
      "epoch": 14.658809428685577,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 36691
    },
    {
      "epoch": 14.659208949260886,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 36692
    },
    {
      "epoch": 14.659608469836197,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 36693
    },
    {
      "epoch": 14.660007990411506,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 36694
    },
    {
      "epoch": 14.660407510986817,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 36695
    },
    {
      "epoch": 14.660807031562126,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 36696
    },
    {
      "epoch": 14.661206552137434,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 36697
    },
    {
      "epoch": 14.661606072712745,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 36698
    },
    {
      "epoch": 14.662005593288054,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 36699
    },
    {
      "epoch": 14.662405113863365,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 36700
    },
    {
      "epoch": 14.662804634438674,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 36701
    },
    {
      "epoch": 14.663204155013982,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 36702
    },
    {
      "epoch": 14.663603675589293,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 36703
    },
    {
      "epoch": 14.664003196164602,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 36704
    },
    {
      "epoch": 14.664402716739913,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 36705
    },
    {
      "epoch": 14.664802237315222,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 36706
    },
    {
      "epoch": 14.665201757890532,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 36707
    },
    {
      "epoch": 14.665601278465841,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 36708
    },
    {
      "epoch": 14.66600079904115,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 36709
    },
    {
      "epoch": 14.66640031961646,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 36710
    },
    {
      "epoch": 14.66679984019177,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 36711
    },
    {
      "epoch": 14.66719936076708,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 36712
    },
    {
      "epoch": 14.66759888134239,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 36713
    },
    {
      "epoch": 14.667998401917698,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 36714
    },
    {
      "epoch": 14.668397922493009,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 36715
    },
    {
      "epoch": 14.668797443068318,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 36716
    },
    {
      "epoch": 14.669196963643628,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0116,
      "step": 36717
    },
    {
      "epoch": 14.669596484218937,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 36718
    },
    {
      "epoch": 14.669996004794246,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 36719
    },
    {
      "epoch": 14.670395525369557,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 36720
    },
    {
      "epoch": 14.670795045944866,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 36721
    },
    {
      "epoch": 14.671194566520176,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 36722
    },
    {
      "epoch": 14.671594087095485,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 36723
    },
    {
      "epoch": 14.671993607670796,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 36724
    },
    {
      "epoch": 14.672393128246105,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 36725
    },
    {
      "epoch": 14.672792648821414,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 36726
    },
    {
      "epoch": 14.673192169396724,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 36727
    },
    {
      "epoch": 14.673591689972033,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 36728
    },
    {
      "epoch": 14.673991210547344,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.005,
      "step": 36729
    },
    {
      "epoch": 14.674390731122653,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 36730
    },
    {
      "epoch": 14.674790251697962,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 36731
    },
    {
      "epoch": 14.675189772273272,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 36732
    },
    {
      "epoch": 14.675589292848581,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 36733
    },
    {
      "epoch": 14.675988813423892,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 36734
    },
    {
      "epoch": 14.676388333999201,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 36735
    },
    {
      "epoch": 14.67678785457451,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 36736
    },
    {
      "epoch": 14.67718737514982,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 36737
    },
    {
      "epoch": 14.67758689572513,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 36738
    },
    {
      "epoch": 14.67798641630044,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 36739
    },
    {
      "epoch": 14.678385936875749,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 36740
    },
    {
      "epoch": 14.67878545745106,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 36741
    },
    {
      "epoch": 14.679184978026369,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 36742
    },
    {
      "epoch": 14.679584498601677,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 36743
    },
    {
      "epoch": 14.679984019176988,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 36744
    },
    {
      "epoch": 14.680383539752297,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 36745
    },
    {
      "epoch": 14.680783060327606,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 36746
    },
    {
      "epoch": 14.681182580902917,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 36747
    },
    {
      "epoch": 14.681582101478226,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 36748
    },
    {
      "epoch": 14.681981622053536,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 36749
    },
    {
      "epoch": 14.682381142628845,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 36750
    },
    {
      "epoch": 14.682780663204156,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 36751
    },
    {
      "epoch": 14.683180183779465,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 36752
    },
    {
      "epoch": 14.683579704354774,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 36753
    },
    {
      "epoch": 14.683979224930084,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 36754
    },
    {
      "epoch": 14.684378745505393,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 36755
    },
    {
      "epoch": 14.684778266080704,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 36756
    },
    {
      "epoch": 14.685177786656013,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 36757
    },
    {
      "epoch": 14.685577307231322,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 36758
    },
    {
      "epoch": 14.685976827806632,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0066,
      "step": 36759
    },
    {
      "epoch": 14.686376348381941,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 36760
    },
    {
      "epoch": 14.686775868957252,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 36761
    },
    {
      "epoch": 14.68717538953256,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 36762
    },
    {
      "epoch": 14.68757491010787,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 36763
    },
    {
      "epoch": 14.68797443068318,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 36764
    },
    {
      "epoch": 14.68837395125849,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 36765
    },
    {
      "epoch": 14.6887734718338,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 36766
    },
    {
      "epoch": 14.689172992409109,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 36767
    },
    {
      "epoch": 14.68957251298442,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 36768
    },
    {
      "epoch": 14.689972033559728,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 36769
    },
    {
      "epoch": 14.690371554135037,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 36770
    },
    {
      "epoch": 14.690771074710348,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 36771
    },
    {
      "epoch": 14.691170595285657,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0037,
      "step": 36772
    },
    {
      "epoch": 14.691570115860968,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 36773
    },
    {
      "epoch": 14.691969636436276,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 36774
    },
    {
      "epoch": 14.692369157011585,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 36775
    },
    {
      "epoch": 14.692768677586896,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 36776
    },
    {
      "epoch": 14.693168198162205,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 36777
    },
    {
      "epoch": 14.693567718737516,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 36778
    },
    {
      "epoch": 14.693967239312824,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 36779
    },
    {
      "epoch": 14.694366759888133,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 36780
    },
    {
      "epoch": 14.694766280463444,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 36781
    },
    {
      "epoch": 14.695165801038753,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 36782
    },
    {
      "epoch": 14.695565321614064,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 36783
    },
    {
      "epoch": 14.695964842189372,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 36784
    },
    {
      "epoch": 14.696364362764683,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 36785
    },
    {
      "epoch": 14.696763883339992,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 36786
    },
    {
      "epoch": 14.697163403915301,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 36787
    },
    {
      "epoch": 14.697562924490612,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 36788
    },
    {
      "epoch": 14.69796244506592,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 36789
    },
    {
      "epoch": 14.698361965641231,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 36790
    },
    {
      "epoch": 14.69876148621654,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 36791
    },
    {
      "epoch": 14.699161006791849,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 36792
    },
    {
      "epoch": 14.69956052736716,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 36793
    },
    {
      "epoch": 14.699960047942469,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 36794
    },
    {
      "epoch": 14.70035956851778,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 36795
    },
    {
      "epoch": 14.700759089093088,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 36796
    },
    {
      "epoch": 14.701158609668397,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 36797
    },
    {
      "epoch": 14.701558130243708,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 36798
    },
    {
      "epoch": 14.701957650819017,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 36799
    },
    {
      "epoch": 14.702357171394327,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 36800
    },
    {
      "epoch": 14.702756691969636,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 36801
    },
    {
      "epoch": 14.703156212544947,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 36802
    },
    {
      "epoch": 14.703555733120256,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 36803
    },
    {
      "epoch": 14.703955253695565,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 36804
    },
    {
      "epoch": 14.704354774270875,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 36805
    },
    {
      "epoch": 14.704754294846184,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0486,
      "step": 36806
    },
    {
      "epoch": 14.705153815421495,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 36807
    },
    {
      "epoch": 14.705553335996804,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 36808
    },
    {
      "epoch": 14.705952856572113,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 36809
    },
    {
      "epoch": 14.706352377147423,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 36810
    },
    {
      "epoch": 14.706751897722732,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 36811
    },
    {
      "epoch": 14.707151418298043,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 36812
    },
    {
      "epoch": 14.707550938873352,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 36813
    },
    {
      "epoch": 14.70795045944866,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 36814
    },
    {
      "epoch": 14.708349980023971,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 36815
    },
    {
      "epoch": 14.70874950059928,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 36816
    },
    {
      "epoch": 14.709149021174591,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 36817
    },
    {
      "epoch": 14.7095485417499,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 36818
    },
    {
      "epoch": 14.70994806232521,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 36819
    },
    {
      "epoch": 14.71034758290052,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 36820
    },
    {
      "epoch": 14.710747103475828,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 36821
    },
    {
      "epoch": 14.711146624051139,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 36822
    },
    {
      "epoch": 14.711546144626448,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 36823
    },
    {
      "epoch": 14.711945665201759,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 36824
    },
    {
      "epoch": 14.712345185777068,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 36825
    },
    {
      "epoch": 14.712744706352376,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 36826
    },
    {
      "epoch": 14.713144226927687,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 36827
    },
    {
      "epoch": 14.713543747502996,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 36828
    },
    {
      "epoch": 14.713943268078307,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 36829
    },
    {
      "epoch": 14.714342788653616,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 36830
    },
    {
      "epoch": 14.714742309228924,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 36831
    },
    {
      "epoch": 14.715141829804235,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 36832
    },
    {
      "epoch": 14.715541350379544,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 36833
    },
    {
      "epoch": 14.715940870954855,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 36834
    },
    {
      "epoch": 14.716340391530164,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 36835
    },
    {
      "epoch": 14.716739912105474,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 36836
    },
    {
      "epoch": 14.717139432680783,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 36837
    },
    {
      "epoch": 14.717538953256092,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 36838
    },
    {
      "epoch": 14.717938473831403,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 36839
    },
    {
      "epoch": 14.718337994406712,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 36840
    },
    {
      "epoch": 14.718737514982022,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 36841
    },
    {
      "epoch": 14.719137035557331,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 36842
    },
    {
      "epoch": 14.71953655613264,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 36843
    },
    {
      "epoch": 14.71993607670795,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 36844
    },
    {
      "epoch": 14.72033559728326,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 36845
    },
    {
      "epoch": 14.72073511785857,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 36846
    },
    {
      "epoch": 14.72113463843388,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 36847
    },
    {
      "epoch": 14.721534159009188,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 36848
    },
    {
      "epoch": 14.721933679584499,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 36849
    },
    {
      "epoch": 14.722333200159808,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 36850
    },
    {
      "epoch": 14.722732720735118,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 36851
    },
    {
      "epoch": 14.723132241310427,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 36852
    },
    {
      "epoch": 14.723531761885738,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 36853
    },
    {
      "epoch": 14.723931282461047,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 36854
    },
    {
      "epoch": 14.724330803036356,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 36855
    },
    {
      "epoch": 14.724730323611666,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 36856
    },
    {
      "epoch": 14.725129844186975,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 36857
    },
    {
      "epoch": 14.725529364762286,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 36858
    },
    {
      "epoch": 14.725928885337595,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 36859
    },
    {
      "epoch": 14.726328405912904,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 36860
    },
    {
      "epoch": 14.726727926488214,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 36861
    },
    {
      "epoch": 14.727127447063523,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 36862
    },
    {
      "epoch": 14.727526967638834,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 36863
    },
    {
      "epoch": 14.727926488214143,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 36864
    },
    {
      "epoch": 14.728326008789452,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 36865
    },
    {
      "epoch": 14.728725529364763,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 36866
    },
    {
      "epoch": 14.729125049940071,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 36867
    },
    {
      "epoch": 14.729524570515382,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 36868
    },
    {
      "epoch": 14.729924091090691,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 36869
    },
    {
      "epoch": 14.730323611666002,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 36870
    },
    {
      "epoch": 14.73072313224131,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 36871
    },
    {
      "epoch": 14.73112265281662,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 36872
    },
    {
      "epoch": 14.73152217339193,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 36873
    },
    {
      "epoch": 14.731921693967239,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 36874
    },
    {
      "epoch": 14.73232121454255,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 36875
    },
    {
      "epoch": 14.732720735117859,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 36876
    },
    {
      "epoch": 14.733120255693168,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 36877
    },
    {
      "epoch": 14.733519776268478,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 36878
    },
    {
      "epoch": 14.733919296843787,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 36879
    },
    {
      "epoch": 14.734318817419098,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 36880
    },
    {
      "epoch": 14.734718337994407,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 36881
    },
    {
      "epoch": 14.735117858569716,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 36882
    },
    {
      "epoch": 14.735517379145026,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 36883
    },
    {
      "epoch": 14.735916899720335,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 36884
    },
    {
      "epoch": 14.736316420295646,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 36885
    },
    {
      "epoch": 14.736715940870955,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 36886
    },
    {
      "epoch": 14.737115461446265,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 36887
    },
    {
      "epoch": 14.737514982021574,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 36888
    },
    {
      "epoch": 14.737914502596883,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 36889
    },
    {
      "epoch": 14.738314023172194,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 36890
    },
    {
      "epoch": 14.738713543747503,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 36891
    },
    {
      "epoch": 14.739113064322813,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 36892
    },
    {
      "epoch": 14.739512584898122,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 36893
    },
    {
      "epoch": 14.739912105473431,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 36894
    },
    {
      "epoch": 14.740311626048742,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 36895
    },
    {
      "epoch": 14.74071114662405,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 36896
    },
    {
      "epoch": 14.741110667199361,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 36897
    },
    {
      "epoch": 14.74151018777467,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 36898
    },
    {
      "epoch": 14.74190970834998,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 36899
    },
    {
      "epoch": 14.74230922892529,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 36900
    },
    {
      "epoch": 14.742708749500599,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 36901
    },
    {
      "epoch": 14.74310827007591,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 36902
    },
    {
      "epoch": 14.743507790651218,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 36903
    },
    {
      "epoch": 14.743907311226529,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 36904
    },
    {
      "epoch": 14.744306831801838,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 36905
    },
    {
      "epoch": 14.744706352377147,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 36906
    },
    {
      "epoch": 14.745105872952458,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 36907
    },
    {
      "epoch": 14.745505393527766,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 36908
    },
    {
      "epoch": 14.745904914103077,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 36909
    },
    {
      "epoch": 14.746304434678386,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 36910
    },
    {
      "epoch": 14.746703955253695,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 36911
    },
    {
      "epoch": 14.747103475829006,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 36912
    },
    {
      "epoch": 14.747502996404315,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 36913
    },
    {
      "epoch": 14.747902516979625,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 36914
    },
    {
      "epoch": 14.748302037554934,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 36915
    },
    {
      "epoch": 14.748701558130243,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 36916
    },
    {
      "epoch": 14.749101078705554,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 36917
    },
    {
      "epoch": 14.749500599280863,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 36918
    },
    {
      "epoch": 14.749900119856173,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 36919
    },
    {
      "epoch": 14.750299640431482,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 36920
    },
    {
      "epoch": 14.750699161006793,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 36921
    },
    {
      "epoch": 14.751098681582102,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0073,
      "step": 36922
    },
    {
      "epoch": 14.75149820215741,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 36923
    },
    {
      "epoch": 14.751897722732721,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 36924
    },
    {
      "epoch": 14.75229724330803,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 36925
    },
    {
      "epoch": 14.752696763883339,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 36926
    },
    {
      "epoch": 14.75309628445865,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 36927
    },
    {
      "epoch": 14.753495805033959,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 36928
    },
    {
      "epoch": 14.75389532560927,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 36929
    },
    {
      "epoch": 14.754294846184578,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 36930
    },
    {
      "epoch": 14.754694366759889,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 36931
    },
    {
      "epoch": 14.755093887335198,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 36932
    },
    {
      "epoch": 14.755493407910507,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 36933
    },
    {
      "epoch": 14.755892928485817,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 36934
    },
    {
      "epoch": 14.756292449061126,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 36935
    },
    {
      "epoch": 14.756691969636437,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 36936
    },
    {
      "epoch": 14.757091490211746,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 36937
    },
    {
      "epoch": 14.757491010787055,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 36938
    },
    {
      "epoch": 14.757890531362365,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 36939
    },
    {
      "epoch": 14.758290051937674,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 36940
    },
    {
      "epoch": 14.758689572512985,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 36941
    },
    {
      "epoch": 14.759089093088294,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 36942
    },
    {
      "epoch": 14.759488613663603,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 36943
    },
    {
      "epoch": 14.759888134238913,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 36944
    },
    {
      "epoch": 14.760287654814222,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 36945
    },
    {
      "epoch": 14.760687175389533,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 36946
    },
    {
      "epoch": 14.761086695964842,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 36947
    },
    {
      "epoch": 14.761486216540153,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 36948
    },
    {
      "epoch": 14.761885737115461,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 36949
    },
    {
      "epoch": 14.76228525769077,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 36950
    },
    {
      "epoch": 14.762684778266081,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 36951
    },
    {
      "epoch": 14.76308429884139,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 36952
    },
    {
      "epoch": 14.7634838194167,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 36953
    },
    {
      "epoch": 14.76388333999201,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 36954
    },
    {
      "epoch": 14.764282860567318,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 36955
    },
    {
      "epoch": 14.764682381142629,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 36956
    },
    {
      "epoch": 14.765081901717938,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 36957
    },
    {
      "epoch": 14.765481422293249,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 36958
    },
    {
      "epoch": 14.765880942868558,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 36959
    },
    {
      "epoch": 14.766280463443866,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 36960
    },
    {
      "epoch": 14.766679984019177,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 36961
    },
    {
      "epoch": 14.767079504594486,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 36962
    },
    {
      "epoch": 14.767479025169797,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 36963
    },
    {
      "epoch": 14.767878545745106,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 36964
    },
    {
      "epoch": 14.768278066320416,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 36965
    },
    {
      "epoch": 14.768677586895725,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 36966
    },
    {
      "epoch": 14.769077107471034,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 36967
    },
    {
      "epoch": 14.769476628046345,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 36968
    },
    {
      "epoch": 14.769876148621654,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 36969
    },
    {
      "epoch": 14.770275669196964,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 36970
    },
    {
      "epoch": 14.770675189772273,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 36971
    },
    {
      "epoch": 14.771074710347582,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 36972
    },
    {
      "epoch": 14.771474230922893,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 36973
    },
    {
      "epoch": 14.771873751498202,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 36974
    },
    {
      "epoch": 14.772273272073512,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 36975
    },
    {
      "epoch": 14.772672792648821,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 36976
    },
    {
      "epoch": 14.77307231322413,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 36977
    },
    {
      "epoch": 14.77347183379944,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 36978
    },
    {
      "epoch": 14.77387135437475,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 36979
    },
    {
      "epoch": 14.77427087495006,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 36980
    },
    {
      "epoch": 14.77467039552537,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 36981
    },
    {
      "epoch": 14.77506991610068,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 36982
    },
    {
      "epoch": 14.775469436675989,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 36983
    },
    {
      "epoch": 14.775868957251298,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 36984
    },
    {
      "epoch": 14.776268477826608,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 36985
    },
    {
      "epoch": 14.776667998401917,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 36986
    },
    {
      "epoch": 14.777067518977228,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 36987
    },
    {
      "epoch": 14.777467039552537,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 36988
    },
    {
      "epoch": 14.777866560127846,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 36989
    },
    {
      "epoch": 14.778266080703157,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 36990
    },
    {
      "epoch": 14.778665601278465,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 36991
    },
    {
      "epoch": 14.779065121853776,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 36992
    },
    {
      "epoch": 14.779464642429085,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 36993
    },
    {
      "epoch": 14.779864163004394,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 36994
    },
    {
      "epoch": 14.780263683579705,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 36995
    },
    {
      "epoch": 14.780663204155013,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 36996
    },
    {
      "epoch": 14.781062724730324,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 36997
    },
    {
      "epoch": 14.781462245305633,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 36998
    },
    {
      "epoch": 14.781861765880944,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 36999
    },
    {
      "epoch": 14.782261286456253,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 37000
    },
    {
      "epoch": 14.782660807031561,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 37001
    },
    {
      "epoch": 14.783060327606872,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 37002
    },
    {
      "epoch": 14.783459848182181,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 37003
    },
    {
      "epoch": 14.783859368757492,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 37004
    },
    {
      "epoch": 14.7842588893328,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 37005
    },
    {
      "epoch": 14.78465840990811,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 37006
    },
    {
      "epoch": 14.78505793048342,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 37007
    },
    {
      "epoch": 14.785457451058729,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 37008
    },
    {
      "epoch": 14.78585697163404,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 37009
    },
    {
      "epoch": 14.786256492209349,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 37010
    },
    {
      "epoch": 14.786656012784658,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 37011
    },
    {
      "epoch": 14.787055533359968,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 37012
    },
    {
      "epoch": 14.787455053935277,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 37013
    },
    {
      "epoch": 14.787854574510588,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 37014
    },
    {
      "epoch": 14.788254095085897,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 37015
    },
    {
      "epoch": 14.788653615661207,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 37016
    },
    {
      "epoch": 14.789053136236516,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 37017
    },
    {
      "epoch": 14.789452656811825,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 37018
    },
    {
      "epoch": 14.789852177387136,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 37019
    },
    {
      "epoch": 14.790251697962445,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 37020
    },
    {
      "epoch": 14.790651218537755,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 37021
    },
    {
      "epoch": 14.791050739113064,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 37022
    },
    {
      "epoch": 14.791450259688373,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 37023
    },
    {
      "epoch": 14.791849780263684,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 37024
    },
    {
      "epoch": 14.792249300838993,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 37025
    },
    {
      "epoch": 14.792648821414303,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 37026
    },
    {
      "epoch": 14.793048341989612,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 37027
    },
    {
      "epoch": 14.793447862564921,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 37028
    },
    {
      "epoch": 14.793847383140232,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 37029
    },
    {
      "epoch": 14.79424690371554,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 37030
    },
    {
      "epoch": 14.794646424290852,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 37031
    },
    {
      "epoch": 14.79504594486616,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 37032
    },
    {
      "epoch": 14.795445465441471,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 37033
    },
    {
      "epoch": 14.79584498601678,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 37034
    },
    {
      "epoch": 14.796244506592089,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 37035
    },
    {
      "epoch": 14.7966440271674,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 37036
    },
    {
      "epoch": 14.797043547742708,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 37037
    },
    {
      "epoch": 14.79744306831802,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 37038
    },
    {
      "epoch": 14.797842588893328,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 37039
    },
    {
      "epoch": 14.798242109468637,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 37040
    },
    {
      "epoch": 14.798641630043948,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 37041
    },
    {
      "epoch": 14.799041150619257,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 37042
    },
    {
      "epoch": 14.799440671194567,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 37043
    },
    {
      "epoch": 14.799840191769876,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 37044
    },
    {
      "epoch": 14.800239712345185,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 37045
    },
    {
      "epoch": 14.800639232920496,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 37046
    },
    {
      "epoch": 14.801038753495805,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 37047
    },
    {
      "epoch": 14.801438274071115,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 37048
    },
    {
      "epoch": 14.801837794646424,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 37049
    },
    {
      "epoch": 14.802237315221735,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 37050
    },
    {
      "epoch": 14.802636835797044,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 37051
    },
    {
      "epoch": 14.803036356372353,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 37052
    },
    {
      "epoch": 14.803435876947663,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 37053
    },
    {
      "epoch": 14.803835397522972,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 37054
    },
    {
      "epoch": 14.804234918098283,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 37055
    },
    {
      "epoch": 14.804634438673592,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 37056
    },
    {
      "epoch": 14.8050339592489,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 37057
    },
    {
      "epoch": 14.805433479824211,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 37058
    },
    {
      "epoch": 14.80583300039952,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 37059
    },
    {
      "epoch": 14.806232520974831,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 37060
    },
    {
      "epoch": 14.80663204155014,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 37061
    },
    {
      "epoch": 14.807031562125449,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 37062
    },
    {
      "epoch": 14.80743108270076,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 37063
    },
    {
      "epoch": 14.807830603276068,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 37064
    },
    {
      "epoch": 14.808230123851379,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 37065
    },
    {
      "epoch": 14.808629644426688,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 37066
    },
    {
      "epoch": 14.809029165001999,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 37067
    },
    {
      "epoch": 14.809428685577307,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 37068
    },
    {
      "epoch": 14.809828206152616,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 37069
    },
    {
      "epoch": 14.810227726727927,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.01,
      "step": 37070
    },
    {
      "epoch": 14.810627247303236,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 37071
    },
    {
      "epoch": 14.811026767878547,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 37072
    },
    {
      "epoch": 14.811426288453855,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 37073
    },
    {
      "epoch": 14.811825809029164,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 37074
    },
    {
      "epoch": 14.812225329604475,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 37075
    },
    {
      "epoch": 14.812624850179784,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 37076
    },
    {
      "epoch": 14.813024370755095,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 37077
    },
    {
      "epoch": 14.813423891330403,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 37078
    },
    {
      "epoch": 14.813823411905712,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 37079
    },
    {
      "epoch": 14.814222932481023,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 37080
    },
    {
      "epoch": 14.814622453056332,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 37081
    },
    {
      "epoch": 14.815021973631643,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 37082
    },
    {
      "epoch": 14.815421494206952,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 37083
    },
    {
      "epoch": 14.815821014782262,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 37084
    },
    {
      "epoch": 14.816220535357571,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 37085
    },
    {
      "epoch": 14.81662005593288,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 37086
    },
    {
      "epoch": 14.81701957650819,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 37087
    },
    {
      "epoch": 14.8174190970835,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 37088
    },
    {
      "epoch": 14.81781861765881,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 37089
    },
    {
      "epoch": 14.81821813823412,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 37090
    },
    {
      "epoch": 14.818617658809428,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 37091
    },
    {
      "epoch": 14.819017179384739,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 37092
    },
    {
      "epoch": 14.819416699960048,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 37093
    },
    {
      "epoch": 14.819816220535358,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 37094
    },
    {
      "epoch": 14.820215741110667,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 37095
    },
    {
      "epoch": 14.820615261685976,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 37096
    },
    {
      "epoch": 14.821014782261287,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 37097
    },
    {
      "epoch": 14.821414302836596,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 37098
    },
    {
      "epoch": 14.821813823411906,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 37099
    },
    {
      "epoch": 14.822213343987215,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 37100
    },
    {
      "epoch": 14.822612864562524,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 37101
    },
    {
      "epoch": 14.823012385137835,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 37102
    },
    {
      "epoch": 14.823411905713144,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 37103
    },
    {
      "epoch": 14.823811426288454,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 37104
    },
    {
      "epoch": 14.824210946863763,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 37105
    },
    {
      "epoch": 14.824610467439072,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 37106
    },
    {
      "epoch": 14.825009988014383,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 37107
    },
    {
      "epoch": 14.825409508589692,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 37108
    },
    {
      "epoch": 14.825809029165002,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 37109
    },
    {
      "epoch": 14.826208549740311,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 37110
    },
    {
      "epoch": 14.826608070315622,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 37111
    },
    {
      "epoch": 14.827007590890931,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 37112
    },
    {
      "epoch": 14.82740711146624,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 37113
    },
    {
      "epoch": 14.82780663204155,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 37114
    },
    {
      "epoch": 14.82820615261686,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 37115
    },
    {
      "epoch": 14.82860567319217,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 37116
    },
    {
      "epoch": 14.829005193767479,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 37117
    },
    {
      "epoch": 14.829404714342788,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 37118
    },
    {
      "epoch": 14.829804234918099,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 37119
    },
    {
      "epoch": 14.830203755493407,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 37120
    },
    {
      "epoch": 14.830603276068718,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 37121
    },
    {
      "epoch": 14.831002796644027,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 37122
    },
    {
      "epoch": 14.831402317219336,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 37123
    },
    {
      "epoch": 14.831801837794647,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 37124
    },
    {
      "epoch": 14.832201358369955,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 37125
    },
    {
      "epoch": 14.832600878945266,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 37126
    },
    {
      "epoch": 14.833000399520575,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 37127
    },
    {
      "epoch": 14.833399920095886,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 37128
    },
    {
      "epoch": 14.833799440671195,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 37129
    },
    {
      "epoch": 14.834198961246504,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 37130
    },
    {
      "epoch": 14.834598481821814,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 37131
    },
    {
      "epoch": 14.834998002397123,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 37132
    },
    {
      "epoch": 14.835397522972434,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 37133
    },
    {
      "epoch": 14.835797043547743,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 37134
    },
    {
      "epoch": 14.836196564123052,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 37135
    },
    {
      "epoch": 14.836596084698362,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 37136
    },
    {
      "epoch": 14.836995605273671,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 37137
    },
    {
      "epoch": 14.837395125848982,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0485,
      "step": 37138
    },
    {
      "epoch": 14.83779464642429,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 37139
    },
    {
      "epoch": 14.8381941669996,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 37140
    },
    {
      "epoch": 14.83859368757491,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 37141
    },
    {
      "epoch": 14.83899320815022,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 37142
    },
    {
      "epoch": 14.83939272872553,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 37143
    },
    {
      "epoch": 14.839792249300839,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 37144
    },
    {
      "epoch": 14.84019176987615,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 37145
    },
    {
      "epoch": 14.840591290451458,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 37146
    },
    {
      "epoch": 14.840990811026767,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 37147
    },
    {
      "epoch": 14.841390331602078,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 37148
    },
    {
      "epoch": 14.841789852177387,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 37149
    },
    {
      "epoch": 14.842189372752697,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 37150
    },
    {
      "epoch": 14.842588893328006,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 37151
    },
    {
      "epoch": 14.842988413903315,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 37152
    },
    {
      "epoch": 14.843387934478626,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 37153
    },
    {
      "epoch": 14.843787455053935,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 37154
    },
    {
      "epoch": 14.844186975629245,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 37155
    },
    {
      "epoch": 14.844586496204554,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 37156
    },
    {
      "epoch": 14.844986016779863,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 37157
    },
    {
      "epoch": 14.845385537355174,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 37158
    },
    {
      "epoch": 14.845785057930483,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 37159
    },
    {
      "epoch": 14.846184578505794,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 37160
    },
    {
      "epoch": 14.846584099081102,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 37161
    },
    {
      "epoch": 14.846983619656413,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 37162
    },
    {
      "epoch": 14.847383140231722,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 37163
    },
    {
      "epoch": 14.847782660807031,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 37164
    },
    {
      "epoch": 14.848182181382342,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 37165
    },
    {
      "epoch": 14.84858170195765,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 37166
    },
    {
      "epoch": 14.848981222532961,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 37167
    },
    {
      "epoch": 14.84938074310827,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 37168
    },
    {
      "epoch": 14.849780263683579,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 37169
    },
    {
      "epoch": 14.85017978425889,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 37170
    },
    {
      "epoch": 14.850579304834199,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 37171
    },
    {
      "epoch": 14.85097882540951,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 37172
    },
    {
      "epoch": 14.851378345984818,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 37173
    },
    {
      "epoch": 14.851777866560127,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 37174
    },
    {
      "epoch": 14.852177387135438,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 37175
    },
    {
      "epoch": 14.852576907710747,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 37176
    },
    {
      "epoch": 14.852976428286057,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 37177
    },
    {
      "epoch": 14.853375948861366,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 37178
    },
    {
      "epoch": 14.853775469436677,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 37179
    },
    {
      "epoch": 14.854174990011986,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 37180
    },
    {
      "epoch": 14.854574510587295,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 37181
    },
    {
      "epoch": 14.854974031162605,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 37182
    },
    {
      "epoch": 14.855373551737914,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 37183
    },
    {
      "epoch": 14.855773072313225,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 37184
    },
    {
      "epoch": 14.856172592888534,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 37185
    },
    {
      "epoch": 14.856572113463843,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 37186
    },
    {
      "epoch": 14.856971634039153,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 37187
    },
    {
      "epoch": 14.857371154614462,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 37188
    },
    {
      "epoch": 14.857770675189773,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 37189
    },
    {
      "epoch": 14.858170195765082,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 37190
    },
    {
      "epoch": 14.85856971634039,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 37191
    },
    {
      "epoch": 14.858969236915701,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 37192
    },
    {
      "epoch": 14.85936875749101,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 37193
    },
    {
      "epoch": 14.859768278066321,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 37194
    },
    {
      "epoch": 14.86016779864163,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 37195
    },
    {
      "epoch": 14.86056731921694,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 37196
    },
    {
      "epoch": 14.86096683979225,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 37197
    },
    {
      "epoch": 14.861366360367558,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 37198
    },
    {
      "epoch": 14.861765880942869,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 37199
    },
    {
      "epoch": 14.862165401518178,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 37200
    },
    {
      "epoch": 14.862564922093489,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 37201
    },
    {
      "epoch": 14.862964442668797,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 37202
    },
    {
      "epoch": 14.863363963244106,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 37203
    },
    {
      "epoch": 14.863763483819417,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 37204
    },
    {
      "epoch": 14.864163004394726,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 37205
    },
    {
      "epoch": 14.864562524970037,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 37206
    },
    {
      "epoch": 14.864962045545346,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 37207
    },
    {
      "epoch": 14.865361566120654,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 37208
    },
    {
      "epoch": 14.865761086695965,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 37209
    },
    {
      "epoch": 14.866160607271274,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 37210
    },
    {
      "epoch": 14.866560127846585,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 37211
    },
    {
      "epoch": 14.866959648421894,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 37212
    },
    {
      "epoch": 14.867359168997204,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 37213
    },
    {
      "epoch": 14.867758689572513,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 37214
    },
    {
      "epoch": 14.868158210147822,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 37215
    },
    {
      "epoch": 14.868557730723133,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 37216
    },
    {
      "epoch": 14.868957251298442,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 37217
    },
    {
      "epoch": 14.869356771873752,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 37218
    },
    {
      "epoch": 14.869756292449061,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 37219
    },
    {
      "epoch": 14.87015581302437,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 37220
    },
    {
      "epoch": 14.87055533359968,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 37221
    },
    {
      "epoch": 14.87095485417499,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 37222
    },
    {
      "epoch": 14.8713543747503,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 37223
    },
    {
      "epoch": 14.87175389532561,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 37224
    },
    {
      "epoch": 14.872153415900918,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 37225
    },
    {
      "epoch": 14.872552936476229,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 37226
    },
    {
      "epoch": 14.872952457051538,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 37227
    },
    {
      "epoch": 14.873351977626848,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 37228
    },
    {
      "epoch": 14.873751498202157,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 37229
    },
    {
      "epoch": 14.874151018777468,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 37230
    },
    {
      "epoch": 14.874550539352777,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 37231
    },
    {
      "epoch": 14.874950059928086,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 37232
    },
    {
      "epoch": 14.875349580503396,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 37233
    },
    {
      "epoch": 14.875749101078705,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 37234
    },
    {
      "epoch": 14.876148621654016,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 37235
    },
    {
      "epoch": 14.876548142229325,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 37236
    },
    {
      "epoch": 14.876947662804634,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 37237
    },
    {
      "epoch": 14.877347183379944,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 37238
    },
    {
      "epoch": 14.877746703955253,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 37239
    },
    {
      "epoch": 14.878146224530564,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 37240
    },
    {
      "epoch": 14.878545745105873,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 37241
    },
    {
      "epoch": 14.878945265681182,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 37242
    },
    {
      "epoch": 14.879344786256492,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 37243
    },
    {
      "epoch": 14.879744306831801,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 37244
    },
    {
      "epoch": 14.880143827407112,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 37245
    },
    {
      "epoch": 14.880543347982421,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 37246
    },
    {
      "epoch": 14.880942868557732,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 37247
    },
    {
      "epoch": 14.88134238913304,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 37248
    },
    {
      "epoch": 14.88174190970835,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 37249
    },
    {
      "epoch": 14.88214143028366,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 37250
    },
    {
      "epoch": 14.882540950858969,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 37251
    },
    {
      "epoch": 14.88294047143428,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 37252
    },
    {
      "epoch": 14.883339992009589,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 37253
    },
    {
      "epoch": 14.883739512584897,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 37254
    },
    {
      "epoch": 14.884139033160208,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 37255
    },
    {
      "epoch": 14.884538553735517,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 37256
    },
    {
      "epoch": 14.884938074310828,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 37257
    },
    {
      "epoch": 14.885337594886137,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 37258
    },
    {
      "epoch": 14.885737115461446,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 37259
    },
    {
      "epoch": 14.886136636036756,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 37260
    },
    {
      "epoch": 14.886536156612065,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 37261
    },
    {
      "epoch": 14.886935677187376,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 37262
    },
    {
      "epoch": 14.887335197762685,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 37263
    },
    {
      "epoch": 14.887734718337995,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 37264
    },
    {
      "epoch": 14.888134238913304,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 37265
    },
    {
      "epoch": 14.888533759488613,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 37266
    },
    {
      "epoch": 14.888933280063924,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 37267
    },
    {
      "epoch": 14.889332800639233,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 37268
    },
    {
      "epoch": 14.889732321214543,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 37269
    },
    {
      "epoch": 14.890131841789852,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 37270
    },
    {
      "epoch": 14.890531362365161,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 37271
    },
    {
      "epoch": 14.890930882940472,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 37272
    },
    {
      "epoch": 14.89133040351578,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 37273
    },
    {
      "epoch": 14.891729924091091,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 37274
    },
    {
      "epoch": 14.8921294446664,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 37275
    },
    {
      "epoch": 14.89252896524171,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 37276
    },
    {
      "epoch": 14.89292848581702,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 37277
    },
    {
      "epoch": 14.893328006392329,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 37278
    },
    {
      "epoch": 14.89372752696764,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 37279
    },
    {
      "epoch": 14.894127047542948,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 37280
    },
    {
      "epoch": 14.894526568118257,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 37281
    },
    {
      "epoch": 14.894926088693568,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 37282
    },
    {
      "epoch": 14.895325609268877,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 37283
    },
    {
      "epoch": 14.895725129844188,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 37284
    },
    {
      "epoch": 14.896124650419496,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 37285
    },
    {
      "epoch": 14.896524170994805,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 37286
    },
    {
      "epoch": 14.896923691570116,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 37287
    },
    {
      "epoch": 14.897323212145425,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 37288
    },
    {
      "epoch": 14.897722732720736,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 37289
    },
    {
      "epoch": 14.898122253296044,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 37290
    },
    {
      "epoch": 14.898521773871355,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 37291
    },
    {
      "epoch": 14.898921294446664,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 37292
    },
    {
      "epoch": 14.899320815021973,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 37293
    },
    {
      "epoch": 14.899720335597284,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 37294
    },
    {
      "epoch": 14.900119856172592,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 37295
    },
    {
      "epoch": 14.900519376747903,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 37296
    },
    {
      "epoch": 14.900918897323212,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 37297
    },
    {
      "epoch": 14.901318417898521,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 37298
    },
    {
      "epoch": 14.901717938473832,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 37299
    },
    {
      "epoch": 14.90211745904914,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 37300
    },
    {
      "epoch": 14.902516979624451,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 37301
    },
    {
      "epoch": 14.90291650019976,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 37302
    },
    {
      "epoch": 14.903316020775069,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 37303
    },
    {
      "epoch": 14.90371554135038,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 37304
    },
    {
      "epoch": 14.904115061925689,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 37305
    },
    {
      "epoch": 14.904514582501,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 37306
    },
    {
      "epoch": 14.904914103076308,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 37307
    },
    {
      "epoch": 14.905313623651619,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 37308
    },
    {
      "epoch": 14.905713144226928,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 37309
    },
    {
      "epoch": 14.906112664802237,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 37310
    },
    {
      "epoch": 14.906512185377547,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 37311
    },
    {
      "epoch": 14.906911705952856,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 37312
    },
    {
      "epoch": 14.907311226528167,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 37313
    },
    {
      "epoch": 14.907710747103476,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 37314
    },
    {
      "epoch": 14.908110267678785,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 37315
    },
    {
      "epoch": 14.908509788254095,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 37316
    },
    {
      "epoch": 14.908909308829404,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 37317
    },
    {
      "epoch": 14.909308829404715,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 37318
    },
    {
      "epoch": 14.909708349980024,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 37319
    },
    {
      "epoch": 14.910107870555333,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 37320
    },
    {
      "epoch": 14.910507391130643,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 37321
    },
    {
      "epoch": 14.910906911705952,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 37322
    },
    {
      "epoch": 14.911306432281263,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 37323
    },
    {
      "epoch": 14.911705952856572,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 37324
    },
    {
      "epoch": 14.912105473431883,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 37325
    },
    {
      "epoch": 14.912504994007191,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 37326
    },
    {
      "epoch": 14.9129045145825,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 37327
    },
    {
      "epoch": 14.913304035157811,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0115,
      "step": 37328
    },
    {
      "epoch": 14.91370355573312,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 37329
    },
    {
      "epoch": 14.91410307630843,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.0455,
      "step": 37330
    },
    {
      "epoch": 14.91450259688374,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 37331
    },
    {
      "epoch": 14.914902117459048,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 37332
    },
    {
      "epoch": 14.915301638034359,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 37333
    },
    {
      "epoch": 14.915701158609668,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 37334
    },
    {
      "epoch": 14.916100679184979,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 37335
    },
    {
      "epoch": 14.916500199760288,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 37336
    },
    {
      "epoch": 14.916899720335596,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 37337
    },
    {
      "epoch": 14.917299240910907,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 37338
    },
    {
      "epoch": 14.917698761486216,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 37339
    },
    {
      "epoch": 14.918098282061527,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 37340
    },
    {
      "epoch": 14.918497802636836,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 37341
    },
    {
      "epoch": 14.918897323212146,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 37342
    },
    {
      "epoch": 14.919296843787455,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 37343
    },
    {
      "epoch": 14.919696364362764,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 37344
    },
    {
      "epoch": 14.920095884938075,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 37345
    },
    {
      "epoch": 14.920495405513384,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 37346
    },
    {
      "epoch": 14.920894926088694,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 37347
    },
    {
      "epoch": 14.921294446664003,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 37348
    },
    {
      "epoch": 14.921693967239312,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 37349
    },
    {
      "epoch": 14.922093487814623,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 37350
    },
    {
      "epoch": 14.922493008389932,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 37351
    },
    {
      "epoch": 14.922892528965242,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 37352
    },
    {
      "epoch": 14.923292049540551,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 37353
    },
    {
      "epoch": 14.92369157011586,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 37354
    },
    {
      "epoch": 14.92409109069117,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 37355
    },
    {
      "epoch": 14.92449061126648,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 37356
    },
    {
      "epoch": 14.92489013184179,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 37357
    },
    {
      "epoch": 14.9252896524171,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 37358
    },
    {
      "epoch": 14.92568917299241,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 37359
    },
    {
      "epoch": 14.926088693567719,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 37360
    },
    {
      "epoch": 14.926488214143028,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 37361
    },
    {
      "epoch": 14.926887734718338,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 37362
    },
    {
      "epoch": 14.927287255293647,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 37363
    },
    {
      "epoch": 14.927686775868958,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 37364
    },
    {
      "epoch": 14.928086296444267,
      "grad_norm": 0.03564453125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 37365
    },
    {
      "epoch": 14.928485817019576,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 37366
    },
    {
      "epoch": 14.928885337594886,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 37367
    },
    {
      "epoch": 14.929284858170195,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 37368
    },
    {
      "epoch": 14.929684378745506,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 37369
    },
    {
      "epoch": 14.930083899320815,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 37370
    },
    {
      "epoch": 14.930483419896124,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 37371
    },
    {
      "epoch": 14.930882940471434,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 37372
    },
    {
      "epoch": 14.931282461046743,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 37373
    },
    {
      "epoch": 14.931681981622054,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 37374
    },
    {
      "epoch": 14.932081502197363,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 37375
    },
    {
      "epoch": 14.932481022772674,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 37376
    },
    {
      "epoch": 14.932880543347983,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 37377
    },
    {
      "epoch": 14.933280063923291,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 37378
    },
    {
      "epoch": 14.933679584498602,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 37379
    },
    {
      "epoch": 14.934079105073911,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 37380
    },
    {
      "epoch": 14.934478625649222,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 37381
    },
    {
      "epoch": 14.93487814622453,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 37382
    },
    {
      "epoch": 14.93527766679984,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 37383
    },
    {
      "epoch": 14.93567718737515,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 37384
    },
    {
      "epoch": 14.936076707950459,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 37385
    },
    {
      "epoch": 14.93647622852577,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 37386
    },
    {
      "epoch": 14.936875749101079,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 37387
    },
    {
      "epoch": 14.937275269676388,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 37388
    },
    {
      "epoch": 14.937674790251698,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 37389
    },
    {
      "epoch": 14.938074310827007,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 37390
    },
    {
      "epoch": 14.938473831402318,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 37391
    },
    {
      "epoch": 14.938873351977627,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 37392
    },
    {
      "epoch": 14.939272872552937,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 37393
    },
    {
      "epoch": 14.939672393128246,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 37394
    },
    {
      "epoch": 14.940071913703555,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 37395
    },
    {
      "epoch": 14.940471434278866,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 37396
    },
    {
      "epoch": 14.940870954854175,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 37397
    },
    {
      "epoch": 14.941270475429485,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 37398
    },
    {
      "epoch": 14.941669996004794,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 37399
    },
    {
      "epoch": 14.942069516580103,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 37400
    },
    {
      "epoch": 14.942469037155414,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 37401
    },
    {
      "epoch": 14.942868557730723,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 37402
    },
    {
      "epoch": 14.943268078306033,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 37403
    },
    {
      "epoch": 14.943667598881342,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 37404
    },
    {
      "epoch": 14.944067119456651,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 37405
    },
    {
      "epoch": 14.944466640031962,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 37406
    },
    {
      "epoch": 14.94486616060727,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 37407
    },
    {
      "epoch": 14.945265681182581,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 37408
    },
    {
      "epoch": 14.94566520175789,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 37409
    },
    {
      "epoch": 14.946064722333201,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 37410
    },
    {
      "epoch": 14.94646424290851,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 37411
    },
    {
      "epoch": 14.946863763483819,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 37412
    },
    {
      "epoch": 14.94726328405913,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 37413
    },
    {
      "epoch": 14.947662804634438,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 37414
    },
    {
      "epoch": 14.948062325209749,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 37415
    },
    {
      "epoch": 14.948461845785058,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 37416
    },
    {
      "epoch": 14.948861366360367,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 37417
    },
    {
      "epoch": 14.949260886935678,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 37418
    },
    {
      "epoch": 14.949660407510986,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 37419
    },
    {
      "epoch": 14.950059928086297,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 37420
    },
    {
      "epoch": 14.950459448661606,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 37421
    },
    {
      "epoch": 14.950858969236915,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 37422
    },
    {
      "epoch": 14.951258489812226,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 37423
    },
    {
      "epoch": 14.951658010387535,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 37424
    },
    {
      "epoch": 14.952057530962845,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 37425
    },
    {
      "epoch": 14.952457051538154,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 37426
    },
    {
      "epoch": 14.952856572113465,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 37427
    },
    {
      "epoch": 14.953256092688774,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 37428
    },
    {
      "epoch": 14.953655613264083,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 37429
    },
    {
      "epoch": 14.954055133839393,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 37430
    },
    {
      "epoch": 14.954454654414702,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 37431
    },
    {
      "epoch": 14.954854174990013,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 37432
    },
    {
      "epoch": 14.955253695565322,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 37433
    },
    {
      "epoch": 14.95565321614063,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 37434
    },
    {
      "epoch": 14.956052736715941,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 37435
    },
    {
      "epoch": 14.95645225729125,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 37436
    },
    {
      "epoch": 14.95685177786656,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 37437
    },
    {
      "epoch": 14.95725129844187,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 37438
    },
    {
      "epoch": 14.957650819017179,
      "grad_norm": 0.03662109375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 37439
    },
    {
      "epoch": 14.95805033959249,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 37440
    },
    {
      "epoch": 14.958449860167798,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 37441
    },
    {
      "epoch": 14.958849380743109,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 37442
    },
    {
      "epoch": 14.959248901318418,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 37443
    },
    {
      "epoch": 14.959648421893728,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 37444
    },
    {
      "epoch": 14.960047942469037,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 37445
    },
    {
      "epoch": 14.960447463044346,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 37446
    },
    {
      "epoch": 14.960846983619657,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 37447
    },
    {
      "epoch": 14.961246504194966,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 37448
    },
    {
      "epoch": 14.961646024770276,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 37449
    },
    {
      "epoch": 14.962045545345585,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 37450
    },
    {
      "epoch": 14.962445065920894,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 37451
    },
    {
      "epoch": 14.962844586496205,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0109,
      "step": 37452
    },
    {
      "epoch": 14.963244107071514,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 37453
    },
    {
      "epoch": 14.963643627646825,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 37454
    },
    {
      "epoch": 14.964043148222133,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 37455
    },
    {
      "epoch": 14.964442668797442,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 37456
    },
    {
      "epoch": 14.964842189372753,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 37457
    },
    {
      "epoch": 14.965241709948062,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 37458
    },
    {
      "epoch": 14.965641230523373,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 37459
    },
    {
      "epoch": 14.966040751098681,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 37460
    },
    {
      "epoch": 14.96644027167399,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 37461
    },
    {
      "epoch": 14.966839792249301,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 37462
    },
    {
      "epoch": 14.96723931282461,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0499,
      "step": 37463
    },
    {
      "epoch": 14.96763883339992,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 37464
    },
    {
      "epoch": 14.96803835397523,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 37465
    },
    {
      "epoch": 14.968437874550538,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 37466
    },
    {
      "epoch": 14.968837395125849,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 37467
    },
    {
      "epoch": 14.969236915701158,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 37468
    },
    {
      "epoch": 14.969636436276469,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 37469
    },
    {
      "epoch": 14.970035956851778,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 37470
    },
    {
      "epoch": 14.970435477427088,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 37471
    },
    {
      "epoch": 14.970834998002397,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 37472
    },
    {
      "epoch": 14.971234518577706,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 37473
    },
    {
      "epoch": 14.971634039153017,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 37474
    },
    {
      "epoch": 14.972033559728326,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 37475
    },
    {
      "epoch": 14.972433080303636,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 37476
    },
    {
      "epoch": 14.972832600878945,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 37477
    },
    {
      "epoch": 14.973232121454254,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 37478
    },
    {
      "epoch": 14.973631642029565,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 37479
    },
    {
      "epoch": 14.974031162604874,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 37480
    },
    {
      "epoch": 14.974430683180184,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 37481
    },
    {
      "epoch": 14.974830203755493,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 37482
    },
    {
      "epoch": 14.975229724330802,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 37483
    },
    {
      "epoch": 14.975629244906113,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 37484
    },
    {
      "epoch": 14.976028765481422,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 37485
    },
    {
      "epoch": 14.976428286056732,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 37486
    },
    {
      "epoch": 14.976827806632041,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 37487
    },
    {
      "epoch": 14.977227327207352,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 37488
    },
    {
      "epoch": 14.97762684778266,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 37489
    },
    {
      "epoch": 14.97802636835797,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 37490
    },
    {
      "epoch": 14.97842588893328,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 37491
    },
    {
      "epoch": 14.97882540950859,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 37492
    },
    {
      "epoch": 14.9792249300839,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 37493
    },
    {
      "epoch": 14.979624450659209,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 37494
    },
    {
      "epoch": 14.980023971234518,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 37495
    },
    {
      "epoch": 14.980423491809828,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 37496
    },
    {
      "epoch": 14.980823012385137,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 37497
    },
    {
      "epoch": 14.981222532960448,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 37498
    },
    {
      "epoch": 14.981622053535757,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 37499
    },
    {
      "epoch": 14.982021574111066,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 37500
    },
    {
      "epoch": 14.982421094686377,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 37501
    },
    {
      "epoch": 14.982820615261685,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 37502
    },
    {
      "epoch": 14.983220135836996,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 37503
    },
    {
      "epoch": 14.983619656412305,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 37504
    },
    {
      "epoch": 14.984019176987616,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 37505
    },
    {
      "epoch": 14.984418697562925,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 37506
    },
    {
      "epoch": 14.984818218138233,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 37507
    },
    {
      "epoch": 14.985217738713544,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 37508
    },
    {
      "epoch": 14.985617259288853,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 37509
    },
    {
      "epoch": 14.986016779864164,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 37510
    },
    {
      "epoch": 14.986416300439473,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 37511
    },
    {
      "epoch": 14.986815821014781,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 37512
    },
    {
      "epoch": 14.987215341590092,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 37513
    },
    {
      "epoch": 14.987614862165401,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 37514
    },
    {
      "epoch": 14.988014382740712,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0104,
      "step": 37515
    },
    {
      "epoch": 14.98841390331602,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 37516
    },
    {
      "epoch": 14.98881342389133,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 37517
    },
    {
      "epoch": 14.98921294446664,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 37518
    },
    {
      "epoch": 14.989612465041949,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 37519
    },
    {
      "epoch": 14.99001198561726,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 37520
    },
    {
      "epoch": 14.990411506192569,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 37521
    },
    {
      "epoch": 14.99081102676788,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 37522
    },
    {
      "epoch": 14.991210547343188,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 37523
    },
    {
      "epoch": 14.991610067918497,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 37524
    },
    {
      "epoch": 14.992009588493808,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 37525
    },
    {
      "epoch": 14.992409109069117,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 37526
    },
    {
      "epoch": 14.992808629644427,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 37527
    },
    {
      "epoch": 14.993208150219736,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 37528
    },
    {
      "epoch": 14.993607670795045,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 37529
    },
    {
      "epoch": 14.994007191370356,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 37530
    },
    {
      "epoch": 14.994406711945665,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 37531
    },
    {
      "epoch": 14.994806232520975,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 37532
    },
    {
      "epoch": 14.995205753096284,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 37533
    },
    {
      "epoch": 14.995605273671593,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 37534
    },
    {
      "epoch": 14.996004794246904,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 37535
    },
    {
      "epoch": 14.996404314822213,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 37536
    },
    {
      "epoch": 14.996803835397523,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 37537
    },
    {
      "epoch": 14.997203355972832,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 37538
    },
    {
      "epoch": 14.997602876548143,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 37539
    },
    {
      "epoch": 14.998002397123452,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 37540
    },
    {
      "epoch": 14.99840191769876,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 37541
    },
    {
      "epoch": 14.998801438274072,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 37542
    },
    {
      "epoch": 14.99920095884938,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 37543
    },
    {
      "epoch": 14.999600479424691,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 37544
    },
    {
      "epoch": 15.0,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 37545
    },
    {
      "epoch": 15.000399520575309,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0114,
      "step": 37546
    },
    {
      "epoch": 15.00079904115062,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0056,
      "step": 37547
    },
    {
      "epoch": 15.001198561725928,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 37548
    },
    {
      "epoch": 15.00159808230124,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 37549
    },
    {
      "epoch": 15.001997602876548,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 37550
    },
    {
      "epoch": 15.002397123451857,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 37551
    },
    {
      "epoch": 15.002796644027168,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 37552
    },
    {
      "epoch": 15.003196164602477,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 37553
    },
    {
      "epoch": 15.003595685177787,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 37554
    },
    {
      "epoch": 15.003995205753096,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 37555
    },
    {
      "epoch": 15.004394726328407,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 37556
    },
    {
      "epoch": 15.004794246903716,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 37557
    },
    {
      "epoch": 15.005193767479025,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 37558
    },
    {
      "epoch": 15.005593288054335,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 37559
    },
    {
      "epoch": 15.005992808629644,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 37560
    },
    {
      "epoch": 15.006392329204955,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 37561
    },
    {
      "epoch": 15.006791849780264,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 37562
    },
    {
      "epoch": 15.007191370355573,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 37563
    },
    {
      "epoch": 15.007590890930883,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 37564
    },
    {
      "epoch": 15.007990411506192,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 37565
    },
    {
      "epoch": 15.008389932081503,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 37566
    },
    {
      "epoch": 15.008789452656812,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 37567
    },
    {
      "epoch": 15.00918897323212,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 37568
    },
    {
      "epoch": 15.009588493807431,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 37569
    },
    {
      "epoch": 15.00998801438274,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 37570
    },
    {
      "epoch": 15.010387534958051,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 37571
    },
    {
      "epoch": 15.01078705553336,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 37572
    },
    {
      "epoch": 15.01118657610867,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 37573
    },
    {
      "epoch": 15.01158609668398,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 37574
    },
    {
      "epoch": 15.011985617259288,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 37575
    },
    {
      "epoch": 15.012385137834599,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 37576
    },
    {
      "epoch": 15.012784658409908,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 37577
    },
    {
      "epoch": 15.013184178985219,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 37578
    },
    {
      "epoch": 15.013583699560527,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 37579
    },
    {
      "epoch": 15.013983220135836,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 37580
    },
    {
      "epoch": 15.014382740711147,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 37581
    },
    {
      "epoch": 15.014782261286456,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 37582
    },
    {
      "epoch": 15.015181781861767,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 37583
    },
    {
      "epoch": 15.015581302437075,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 37584
    },
    {
      "epoch": 15.015980823012384,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 37585
    },
    {
      "epoch": 15.016380343587695,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 37586
    },
    {
      "epoch": 15.016779864163004,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 37587
    },
    {
      "epoch": 15.017179384738315,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 37588
    },
    {
      "epoch": 15.017578905313623,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0135,
      "step": 37589
    },
    {
      "epoch": 15.017978425888934,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 37590
    },
    {
      "epoch": 15.018377946464243,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 37591
    },
    {
      "epoch": 15.018777467039552,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 37592
    },
    {
      "epoch": 15.019176987614863,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 37593
    },
    {
      "epoch": 15.019576508190172,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 37594
    },
    {
      "epoch": 15.019976028765482,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 37595
    },
    {
      "epoch": 15.020375549340791,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 37596
    },
    {
      "epoch": 15.0207750699161,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 37597
    },
    {
      "epoch": 15.02117459049141,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 37598
    },
    {
      "epoch": 15.02157411106672,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 37599
    },
    {
      "epoch": 15.02197363164203,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 37600
    },
    {
      "epoch": 15.02237315221734,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 37601
    },
    {
      "epoch": 15.022772672792648,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 37602
    },
    {
      "epoch": 15.023172193367959,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 37603
    },
    {
      "epoch": 15.023571713943268,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 37604
    },
    {
      "epoch": 15.023971234518578,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 37605
    },
    {
      "epoch": 15.024370755093887,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 37606
    },
    {
      "epoch": 15.024770275669196,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 37607
    },
    {
      "epoch": 15.025169796244507,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 37608
    },
    {
      "epoch": 15.025569316819816,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0101,
      "step": 37609
    },
    {
      "epoch": 15.025968837395126,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 37610
    },
    {
      "epoch": 15.026368357970435,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 37611
    },
    {
      "epoch": 15.026767878545746,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 37612
    },
    {
      "epoch": 15.027167399121055,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 37613
    },
    {
      "epoch": 15.027566919696364,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 37614
    },
    {
      "epoch": 15.027966440271674,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 37615
    },
    {
      "epoch": 15.028365960846983,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 37616
    },
    {
      "epoch": 15.028765481422294,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 37617
    },
    {
      "epoch": 15.029165001997603,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 37618
    },
    {
      "epoch": 15.029564522572912,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 37619
    },
    {
      "epoch": 15.029964043148222,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 37620
    },
    {
      "epoch": 15.030363563723531,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 37621
    },
    {
      "epoch": 15.030763084298842,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 37622
    },
    {
      "epoch": 15.031162604874151,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 37623
    },
    {
      "epoch": 15.03156212544946,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 37624
    },
    {
      "epoch": 15.03196164602477,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 37625
    },
    {
      "epoch": 15.03236116660008,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 37626
    },
    {
      "epoch": 15.03276068717539,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 37627
    },
    {
      "epoch": 15.033160207750699,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 37628
    },
    {
      "epoch": 15.03355972832601,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 37629
    },
    {
      "epoch": 15.033959248901319,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0472,
      "step": 37630
    },
    {
      "epoch": 15.034358769476627,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 37631
    },
    {
      "epoch": 15.034758290051938,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0099,
      "step": 37632
    },
    {
      "epoch": 15.035157810627247,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 37633
    },
    {
      "epoch": 15.035557331202558,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 37634
    },
    {
      "epoch": 15.035956851777867,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 37635
    },
    {
      "epoch": 15.036356372353175,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 37636
    },
    {
      "epoch": 15.036755892928486,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 37637
    },
    {
      "epoch": 15.037155413503795,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 37638
    },
    {
      "epoch": 15.037554934079106,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 37639
    },
    {
      "epoch": 15.037954454654415,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 37640
    },
    {
      "epoch": 15.038353975229724,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 37641
    },
    {
      "epoch": 15.038753495805034,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 37642
    },
    {
      "epoch": 15.039153016380343,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 37643
    },
    {
      "epoch": 15.039552536955654,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 37644
    },
    {
      "epoch": 15.039952057530963,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 37645
    },
    {
      "epoch": 15.040351578106273,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 37646
    },
    {
      "epoch": 15.040751098681582,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 37647
    },
    {
      "epoch": 15.041150619256891,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 37648
    },
    {
      "epoch": 15.041550139832202,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0114,
      "step": 37649
    },
    {
      "epoch": 15.04194966040751,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0107,
      "step": 37650
    },
    {
      "epoch": 15.042349180982821,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 37651
    },
    {
      "epoch": 15.04274870155813,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 37652
    },
    {
      "epoch": 15.04314822213344,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 37653
    },
    {
      "epoch": 15.04354774270875,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 37654
    },
    {
      "epoch": 15.043947263284059,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 37655
    },
    {
      "epoch": 15.04434678385937,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 37656
    },
    {
      "epoch": 15.044746304434678,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 37657
    },
    {
      "epoch": 15.045145825009987,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 37658
    },
    {
      "epoch": 15.045545345585298,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 37659
    },
    {
      "epoch": 15.045944866160607,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 37660
    },
    {
      "epoch": 15.046344386735917,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 37661
    },
    {
      "epoch": 15.046743907311226,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 37662
    },
    {
      "epoch": 15.047143427886537,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 37663
    },
    {
      "epoch": 15.047542948461846,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 37664
    },
    {
      "epoch": 15.047942469037155,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 37665
    },
    {
      "epoch": 15.048341989612465,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 37666
    },
    {
      "epoch": 15.048741510187774,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 37667
    },
    {
      "epoch": 15.049141030763085,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 37668
    },
    {
      "epoch": 15.049540551338394,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 37669
    },
    {
      "epoch": 15.049940071913703,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 37670
    },
    {
      "epoch": 15.050339592489014,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 37671
    },
    {
      "epoch": 15.050739113064322,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 37672
    },
    {
      "epoch": 15.051138633639633,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 37673
    },
    {
      "epoch": 15.051538154214942,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 37674
    },
    {
      "epoch": 15.051937674790251,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 37675
    },
    {
      "epoch": 15.052337195365562,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 37676
    },
    {
      "epoch": 15.05273671594087,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 37677
    },
    {
      "epoch": 15.053136236516181,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 37678
    },
    {
      "epoch": 15.05353575709149,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 37679
    },
    {
      "epoch": 15.053935277666799,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 37680
    },
    {
      "epoch": 15.05433479824211,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 37681
    },
    {
      "epoch": 15.054734318817419,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 37682
    },
    {
      "epoch": 15.05513383939273,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 37683
    },
    {
      "epoch": 15.055533359968038,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 37684
    },
    {
      "epoch": 15.055932880543349,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 37685
    },
    {
      "epoch": 15.056332401118658,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 37686
    },
    {
      "epoch": 15.056731921693967,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 37687
    },
    {
      "epoch": 15.057131442269277,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 37688
    },
    {
      "epoch": 15.057530962844586,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 37689
    },
    {
      "epoch": 15.057930483419897,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 37690
    },
    {
      "epoch": 15.058330003995206,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 37691
    },
    {
      "epoch": 15.058729524570515,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 37692
    },
    {
      "epoch": 15.059129045145825,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 37693
    },
    {
      "epoch": 15.059528565721134,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 37694
    },
    {
      "epoch": 15.059928086296445,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 37695
    },
    {
      "epoch": 15.060327606871754,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 37696
    },
    {
      "epoch": 15.060727127447063,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 37697
    },
    {
      "epoch": 15.061126648022373,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 37698
    },
    {
      "epoch": 15.061526168597682,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 37699
    },
    {
      "epoch": 15.061925689172993,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 37700
    },
    {
      "epoch": 15.062325209748302,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 37701
    },
    {
      "epoch": 15.062724730323612,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 37702
    },
    {
      "epoch": 15.063124250898921,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 37703
    },
    {
      "epoch": 15.06352377147423,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 37704
    },
    {
      "epoch": 15.063923292049541,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 37705
    },
    {
      "epoch": 15.06432281262485,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0124,
      "step": 37706
    },
    {
      "epoch": 15.06472233320016,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 37707
    },
    {
      "epoch": 15.06512185377547,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 37708
    },
    {
      "epoch": 15.065521374350778,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0098,
      "step": 37709
    },
    {
      "epoch": 15.065920894926089,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 37710
    },
    {
      "epoch": 15.066320415501398,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 37711
    },
    {
      "epoch": 15.066719936076709,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 37712
    },
    {
      "epoch": 15.067119456652017,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 37713
    },
    {
      "epoch": 15.067518977227326,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 37714
    },
    {
      "epoch": 15.067918497802637,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 37715
    },
    {
      "epoch": 15.068318018377946,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 37716
    },
    {
      "epoch": 15.068717538953257,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 37717
    },
    {
      "epoch": 15.069117059528566,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 37718
    },
    {
      "epoch": 15.069516580103876,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 37719
    },
    {
      "epoch": 15.069916100679185,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 37720
    },
    {
      "epoch": 15.070315621254494,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 37721
    },
    {
      "epoch": 15.070715141829805,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 37722
    },
    {
      "epoch": 15.071114662405114,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 37723
    },
    {
      "epoch": 15.071514182980424,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 37724
    },
    {
      "epoch": 15.071913703555733,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 37725
    },
    {
      "epoch": 15.072313224131042,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 37726
    },
    {
      "epoch": 15.072712744706353,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 37727
    },
    {
      "epoch": 15.073112265281662,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 37728
    },
    {
      "epoch": 15.073511785856972,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 37729
    },
    {
      "epoch": 15.073911306432281,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 37730
    },
    {
      "epoch": 15.07431082700759,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 37731
    },
    {
      "epoch": 15.0747103475829,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 37732
    },
    {
      "epoch": 15.07510986815821,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 37733
    },
    {
      "epoch": 15.07550938873352,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 37734
    },
    {
      "epoch": 15.07590890930883,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 37735
    },
    {
      "epoch": 15.07630842988414,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 37736
    },
    {
      "epoch": 15.076707950459449,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 37737
    },
    {
      "epoch": 15.077107471034758,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 37738
    },
    {
      "epoch": 15.077506991610068,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 37739
    },
    {
      "epoch": 15.077906512185377,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 37740
    },
    {
      "epoch": 15.078306032760688,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 37741
    },
    {
      "epoch": 15.078705553335997,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 37742
    },
    {
      "epoch": 15.079105073911306,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 37743
    },
    {
      "epoch": 15.079504594486616,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 37744
    },
    {
      "epoch": 15.079904115061925,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 37745
    },
    {
      "epoch": 15.080303635637236,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 37746
    },
    {
      "epoch": 15.080703156212545,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 37747
    },
    {
      "epoch": 15.081102676787854,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 37748
    },
    {
      "epoch": 15.081502197363164,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 37749
    },
    {
      "epoch": 15.081901717938473,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 37750
    },
    {
      "epoch": 15.082301238513784,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 37751
    },
    {
      "epoch": 15.082700759089093,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 37752
    },
    {
      "epoch": 15.083100279664404,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0483,
      "step": 37753
    },
    {
      "epoch": 15.083499800239712,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 37754
    },
    {
      "epoch": 15.083899320815021,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 37755
    },
    {
      "epoch": 15.084298841390332,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 37756
    },
    {
      "epoch": 15.084698361965641,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 37757
    },
    {
      "epoch": 15.085097882540952,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 37758
    },
    {
      "epoch": 15.08549740311626,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 37759
    },
    {
      "epoch": 15.08589692369157,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 37760
    },
    {
      "epoch": 15.08629644426688,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 37761
    },
    {
      "epoch": 15.086695964842189,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 37762
    },
    {
      "epoch": 15.0870954854175,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 37763
    },
    {
      "epoch": 15.087495005992809,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 37764
    },
    {
      "epoch": 15.087894526568117,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 37765
    },
    {
      "epoch": 15.088294047143428,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 37766
    },
    {
      "epoch": 15.088693567718737,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 37767
    },
    {
      "epoch": 15.089093088294048,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 37768
    },
    {
      "epoch": 15.089492608869357,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 37769
    },
    {
      "epoch": 15.089892129444667,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 37770
    },
    {
      "epoch": 15.090291650019976,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 37771
    },
    {
      "epoch": 15.090691170595285,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 37772
    },
    {
      "epoch": 15.091090691170596,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 37773
    },
    {
      "epoch": 15.091490211745905,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 37774
    },
    {
      "epoch": 15.091889732321215,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 37775
    },
    {
      "epoch": 15.092289252896524,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 37776
    },
    {
      "epoch": 15.092688773471833,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 37777
    },
    {
      "epoch": 15.093088294047144,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 37778
    },
    {
      "epoch": 15.093487814622453,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 37779
    },
    {
      "epoch": 15.093887335197763,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 37780
    },
    {
      "epoch": 15.094286855773072,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 37781
    },
    {
      "epoch": 15.094686376348381,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 37782
    },
    {
      "epoch": 15.095085896923692,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 37783
    },
    {
      "epoch": 15.095485417499,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 37784
    },
    {
      "epoch": 15.095884938074311,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 37785
    },
    {
      "epoch": 15.09628445864962,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 37786
    },
    {
      "epoch": 15.09668397922493,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 37787
    },
    {
      "epoch": 15.09708349980024,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 37788
    },
    {
      "epoch": 15.097483020375549,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 37789
    },
    {
      "epoch": 15.09788254095086,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 37790
    },
    {
      "epoch": 15.098282061526168,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 37791
    },
    {
      "epoch": 15.098681582101479,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 37792
    },
    {
      "epoch": 15.099081102676788,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 37793
    },
    {
      "epoch": 15.099480623252097,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 37794
    },
    {
      "epoch": 15.099880143827408,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 37795
    },
    {
      "epoch": 15.100279664402716,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 37796
    },
    {
      "epoch": 15.100679184978027,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 37797
    },
    {
      "epoch": 15.101078705553336,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 37798
    },
    {
      "epoch": 15.101478226128645,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 37799
    },
    {
      "epoch": 15.101877746703956,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 37800
    },
    {
      "epoch": 15.102277267279264,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 37801
    },
    {
      "epoch": 15.102676787854575,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 37802
    },
    {
      "epoch": 15.103076308429884,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 37803
    },
    {
      "epoch": 15.103475829005193,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 37804
    },
    {
      "epoch": 15.103875349580504,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 37805
    },
    {
      "epoch": 15.104274870155812,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 37806
    },
    {
      "epoch": 15.104674390731123,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 37807
    },
    {
      "epoch": 15.105073911306432,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 37808
    },
    {
      "epoch": 15.105473431881743,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 37809
    },
    {
      "epoch": 15.105872952457052,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 37810
    },
    {
      "epoch": 15.10627247303236,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 37811
    },
    {
      "epoch": 15.106671993607671,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 37812
    },
    {
      "epoch": 15.10707151418298,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 37813
    },
    {
      "epoch": 15.10747103475829,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 37814
    },
    {
      "epoch": 15.1078705553336,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0115,
      "step": 37815
    },
    {
      "epoch": 15.108270075908909,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 37816
    },
    {
      "epoch": 15.10866959648422,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 37817
    },
    {
      "epoch": 15.109069117059528,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 37818
    },
    {
      "epoch": 15.109468637634839,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 37819
    },
    {
      "epoch": 15.109868158210148,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 37820
    },
    {
      "epoch": 15.110267678785457,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 37821
    },
    {
      "epoch": 15.110667199360767,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 37822
    },
    {
      "epoch": 15.111066719936076,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 37823
    },
    {
      "epoch": 15.111466240511387,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 37824
    },
    {
      "epoch": 15.111865761086696,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 37825
    },
    {
      "epoch": 15.112265281662006,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 37826
    },
    {
      "epoch": 15.112664802237315,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 37827
    },
    {
      "epoch": 15.113064322812624,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 37828
    },
    {
      "epoch": 15.113463843387935,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 37829
    },
    {
      "epoch": 15.113863363963244,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 37830
    },
    {
      "epoch": 15.114262884538554,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 37831
    },
    {
      "epoch": 15.114662405113863,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 37832
    },
    {
      "epoch": 15.115061925689172,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 37833
    },
    {
      "epoch": 15.115461446264483,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 37834
    },
    {
      "epoch": 15.115860966839792,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 37835
    },
    {
      "epoch": 15.116260487415103,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 37836
    },
    {
      "epoch": 15.116660007990411,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 37837
    },
    {
      "epoch": 15.11705952856572,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 37838
    },
    {
      "epoch": 15.117459049141031,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 37839
    },
    {
      "epoch": 15.11785856971634,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 37840
    },
    {
      "epoch": 15.11825809029165,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 37841
    },
    {
      "epoch": 15.11865761086696,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 37842
    },
    {
      "epoch": 15.11905713144227,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 37843
    },
    {
      "epoch": 15.119456652017579,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 37844
    },
    {
      "epoch": 15.119856172592888,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 37845
    },
    {
      "epoch": 15.120255693168199,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 37846
    },
    {
      "epoch": 15.120655213743508,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 37847
    },
    {
      "epoch": 15.121054734318818,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 37848
    },
    {
      "epoch": 15.121454254894127,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 37849
    },
    {
      "epoch": 15.121853775469436,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 37850
    },
    {
      "epoch": 15.122253296044747,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 37851
    },
    {
      "epoch": 15.122652816620056,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 37852
    },
    {
      "epoch": 15.123052337195366,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 37853
    },
    {
      "epoch": 15.123451857770675,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0128,
      "step": 37854
    },
    {
      "epoch": 15.123851378345984,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 37855
    },
    {
      "epoch": 15.124250898921295,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 37856
    },
    {
      "epoch": 15.124650419496604,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 37857
    },
    {
      "epoch": 15.125049940071914,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 37858
    },
    {
      "epoch": 15.125449460647223,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 37859
    },
    {
      "epoch": 15.125848981222532,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 37860
    },
    {
      "epoch": 15.126248501797843,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 37861
    },
    {
      "epoch": 15.126648022373152,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0024,
      "step": 37862
    },
    {
      "epoch": 15.127047542948462,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 37863
    },
    {
      "epoch": 15.127447063523771,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 37864
    },
    {
      "epoch": 15.127846584099082,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 37865
    },
    {
      "epoch": 15.12824610467439,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 37866
    },
    {
      "epoch": 15.1286456252497,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 37867
    },
    {
      "epoch": 15.12904514582501,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 37868
    },
    {
      "epoch": 15.12944466640032,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 37869
    },
    {
      "epoch": 15.12984418697563,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 37870
    },
    {
      "epoch": 15.130243707550939,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 37871
    },
    {
      "epoch": 15.130643228126248,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 37872
    },
    {
      "epoch": 15.131042748701558,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 37873
    },
    {
      "epoch": 15.131442269276867,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 37874
    },
    {
      "epoch": 15.131841789852178,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 37875
    },
    {
      "epoch": 15.132241310427487,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 37876
    },
    {
      "epoch": 15.132640831002796,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 37877
    },
    {
      "epoch": 15.133040351578106,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 37878
    },
    {
      "epoch": 15.133439872153415,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 37879
    },
    {
      "epoch": 15.133839392728726,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 37880
    },
    {
      "epoch": 15.134238913304035,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 37881
    },
    {
      "epoch": 15.134638433879346,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 37882
    },
    {
      "epoch": 15.135037954454654,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 37883
    },
    {
      "epoch": 15.135437475029963,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 37884
    },
    {
      "epoch": 15.135836995605274,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 37885
    },
    {
      "epoch": 15.136236516180583,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 37886
    },
    {
      "epoch": 15.136636036755894,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 37887
    },
    {
      "epoch": 15.137035557331203,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 37888
    },
    {
      "epoch": 15.137435077906511,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 37889
    },
    {
      "epoch": 15.137834598481822,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 37890
    },
    {
      "epoch": 15.138234119057131,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 37891
    },
    {
      "epoch": 15.138633639632442,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 37892
    },
    {
      "epoch": 15.13903316020775,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 37893
    },
    {
      "epoch": 15.13943268078306,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 37894
    },
    {
      "epoch": 15.13983220135837,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 37895
    },
    {
      "epoch": 15.140231721933679,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 37896
    },
    {
      "epoch": 15.14063124250899,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 37897
    },
    {
      "epoch": 15.141030763084299,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 37898
    },
    {
      "epoch": 15.14143028365961,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 37899
    },
    {
      "epoch": 15.141829804234918,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 37900
    },
    {
      "epoch": 15.142229324810227,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 37901
    },
    {
      "epoch": 15.142628845385538,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 37902
    },
    {
      "epoch": 15.143028365960847,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 37903
    },
    {
      "epoch": 15.143427886536157,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 37904
    },
    {
      "epoch": 15.143827407111466,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 37905
    },
    {
      "epoch": 15.144226927686775,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 37906
    },
    {
      "epoch": 15.144626448262086,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 37907
    },
    {
      "epoch": 15.145025968837395,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 37908
    },
    {
      "epoch": 15.145425489412705,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 37909
    },
    {
      "epoch": 15.145825009988014,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 37910
    },
    {
      "epoch": 15.146224530563323,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 37911
    },
    {
      "epoch": 15.146624051138634,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 37912
    },
    {
      "epoch": 15.147023571713943,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 37913
    },
    {
      "epoch": 15.147423092289253,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 37914
    },
    {
      "epoch": 15.147822612864562,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 37915
    },
    {
      "epoch": 15.148222133439873,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 37916
    },
    {
      "epoch": 15.148621654015182,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 37917
    },
    {
      "epoch": 15.14902117459049,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 37918
    },
    {
      "epoch": 15.149420695165801,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 37919
    },
    {
      "epoch": 15.14982021574111,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 37920
    },
    {
      "epoch": 15.150219736316421,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 37921
    },
    {
      "epoch": 15.15061925689173,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 37922
    },
    {
      "epoch": 15.151018777467039,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 37923
    },
    {
      "epoch": 15.15141829804235,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 37924
    },
    {
      "epoch": 15.151817818617658,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 37925
    },
    {
      "epoch": 15.152217339192969,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 37926
    },
    {
      "epoch": 15.152616859768278,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 37927
    },
    {
      "epoch": 15.153016380343587,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 37928
    },
    {
      "epoch": 15.153415900918898,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 37929
    },
    {
      "epoch": 15.153815421494206,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 37930
    },
    {
      "epoch": 15.154214942069517,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 37931
    },
    {
      "epoch": 15.154614462644826,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 37932
    },
    {
      "epoch": 15.155013983220137,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 37933
    },
    {
      "epoch": 15.155413503795446,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 37934
    },
    {
      "epoch": 15.155813024370755,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 37935
    },
    {
      "epoch": 15.156212544946065,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 37936
    },
    {
      "epoch": 15.156612065521374,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 37937
    },
    {
      "epoch": 15.157011586096685,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 37938
    },
    {
      "epoch": 15.157411106671994,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 37939
    },
    {
      "epoch": 15.157810627247303,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 37940
    },
    {
      "epoch": 15.158210147822613,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 37941
    },
    {
      "epoch": 15.158609668397922,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0095,
      "step": 37942
    },
    {
      "epoch": 15.159009188973233,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 37943
    },
    {
      "epoch": 15.159408709548542,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 37944
    },
    {
      "epoch": 15.15980823012385,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 37945
    },
    {
      "epoch": 15.160207750699161,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 37946
    },
    {
      "epoch": 15.16060727127447,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 37947
    },
    {
      "epoch": 15.16100679184978,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 37948
    },
    {
      "epoch": 15.16140631242509,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 37949
    },
    {
      "epoch": 15.1618058330004,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 37950
    },
    {
      "epoch": 15.16220535357571,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 37951
    },
    {
      "epoch": 15.162604874151018,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 37952
    },
    {
      "epoch": 15.163004394726329,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 37953
    },
    {
      "epoch": 15.163403915301638,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 37954
    },
    {
      "epoch": 15.163803435876948,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 37955
    },
    {
      "epoch": 15.164202956452257,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 37956
    },
    {
      "epoch": 15.164602477027566,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 37957
    },
    {
      "epoch": 15.165001997602877,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 37958
    },
    {
      "epoch": 15.165401518178186,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 37959
    },
    {
      "epoch": 15.165801038753496,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 37960
    },
    {
      "epoch": 15.166200559328805,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 37961
    },
    {
      "epoch": 15.166600079904114,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 37962
    },
    {
      "epoch": 15.166999600479425,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 37963
    },
    {
      "epoch": 15.167399121054734,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 37964
    },
    {
      "epoch": 15.167798641630045,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 37965
    },
    {
      "epoch": 15.168198162205353,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 37966
    },
    {
      "epoch": 15.168597682780662,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 37967
    },
    {
      "epoch": 15.168997203355973,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 37968
    },
    {
      "epoch": 15.169396723931282,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 37969
    },
    {
      "epoch": 15.169796244506593,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 37970
    },
    {
      "epoch": 15.170195765081901,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 37971
    },
    {
      "epoch": 15.170595285657212,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 37972
    },
    {
      "epoch": 15.170994806232521,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 37973
    },
    {
      "epoch": 15.17139432680783,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 37974
    },
    {
      "epoch": 15.17179384738314,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 37975
    },
    {
      "epoch": 15.17219336795845,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 37976
    },
    {
      "epoch": 15.17259288853376,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 37977
    },
    {
      "epoch": 15.172992409109069,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 37978
    },
    {
      "epoch": 15.173391929684378,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 37979
    },
    {
      "epoch": 15.173791450259689,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 37980
    },
    {
      "epoch": 15.174190970834998,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 37981
    },
    {
      "epoch": 15.174590491410308,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 37982
    },
    {
      "epoch": 15.174990011985617,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 37983
    },
    {
      "epoch": 15.175389532560926,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 37984
    },
    {
      "epoch": 15.175789053136237,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 37985
    },
    {
      "epoch": 15.176188573711546,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 37986
    },
    {
      "epoch": 15.176588094286856,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 37987
    },
    {
      "epoch": 15.176987614862165,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 37988
    },
    {
      "epoch": 15.177387135437476,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 37989
    },
    {
      "epoch": 15.177786656012785,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 37990
    },
    {
      "epoch": 15.178186176588094,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 37991
    },
    {
      "epoch": 15.178585697163404,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 37992
    },
    {
      "epoch": 15.178985217738713,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 37993
    },
    {
      "epoch": 15.179384738314024,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 37994
    },
    {
      "epoch": 15.179784258889333,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 37995
    },
    {
      "epoch": 15.180183779464642,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 37996
    },
    {
      "epoch": 15.180583300039952,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 37997
    },
    {
      "epoch": 15.180982820615261,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 37998
    },
    {
      "epoch": 15.181382341190572,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 37999
    },
    {
      "epoch": 15.18178186176588,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 38000
    },
    {
      "epoch": 15.18218138234119,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 38001
    },
    {
      "epoch": 15.1825809029165,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 38002
    },
    {
      "epoch": 15.18298042349181,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 38003
    },
    {
      "epoch": 15.18337994406712,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 38004
    },
    {
      "epoch": 15.183779464642429,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 38005
    },
    {
      "epoch": 15.18417898521774,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 38006
    },
    {
      "epoch": 15.184578505793048,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 38007
    },
    {
      "epoch": 15.184978026368357,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 38008
    },
    {
      "epoch": 15.185377546943668,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 38009
    },
    {
      "epoch": 15.185777067518977,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 38010
    },
    {
      "epoch": 15.186176588094288,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 38011
    },
    {
      "epoch": 15.186576108669597,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 38012
    },
    {
      "epoch": 15.186975629244905,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 38013
    },
    {
      "epoch": 15.187375149820216,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 38014
    },
    {
      "epoch": 15.187774670395525,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 38015
    },
    {
      "epoch": 15.188174190970836,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 38016
    },
    {
      "epoch": 15.188573711546145,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 38017
    },
    {
      "epoch": 15.188973232121453,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 38018
    },
    {
      "epoch": 15.189372752696764,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 38019
    },
    {
      "epoch": 15.189772273272073,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 38020
    },
    {
      "epoch": 15.190171793847384,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 38021
    },
    {
      "epoch": 15.190571314422693,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 38022
    },
    {
      "epoch": 15.190970834998003,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 38023
    },
    {
      "epoch": 15.191370355573312,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 38024
    },
    {
      "epoch": 15.191769876148621,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 38025
    },
    {
      "epoch": 15.192169396723932,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 38026
    },
    {
      "epoch": 15.19256891729924,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 38027
    },
    {
      "epoch": 15.192968437874551,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 38028
    },
    {
      "epoch": 15.19336795844986,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 38029
    },
    {
      "epoch": 15.193767479025169,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 38030
    },
    {
      "epoch": 15.19416699960048,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 38031
    },
    {
      "epoch": 15.194566520175789,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 38032
    },
    {
      "epoch": 15.1949660407511,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 38033
    },
    {
      "epoch": 15.195365561326408,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 38034
    },
    {
      "epoch": 15.195765081901717,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 38035
    },
    {
      "epoch": 15.196164602477028,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 38036
    },
    {
      "epoch": 15.196564123052337,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 38037
    },
    {
      "epoch": 15.196963643627647,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 38038
    },
    {
      "epoch": 15.197363164202956,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 38039
    },
    {
      "epoch": 15.197762684778265,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 38040
    },
    {
      "epoch": 15.198162205353576,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 38041
    },
    {
      "epoch": 15.198561725928885,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 38042
    },
    {
      "epoch": 15.198961246504195,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 38043
    },
    {
      "epoch": 15.199360767079504,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 38044
    },
    {
      "epoch": 15.199760287654815,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 38045
    },
    {
      "epoch": 15.200159808230124,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 38046
    },
    {
      "epoch": 15.200559328805433,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 38047
    },
    {
      "epoch": 15.200958849380743,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 38048
    },
    {
      "epoch": 15.201358369956052,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 38049
    },
    {
      "epoch": 15.201757890531363,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 38050
    },
    {
      "epoch": 15.202157411106672,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 38051
    },
    {
      "epoch": 15.20255693168198,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 38052
    },
    {
      "epoch": 15.202956452257292,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 38053
    },
    {
      "epoch": 15.2033559728326,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 38054
    },
    {
      "epoch": 15.203755493407911,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0116,
      "step": 38055
    },
    {
      "epoch": 15.20415501398322,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 38056
    },
    {
      "epoch": 15.204554534558529,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 38057
    },
    {
      "epoch": 15.20495405513384,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 38058
    },
    {
      "epoch": 15.205353575709148,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 38059
    },
    {
      "epoch": 15.20575309628446,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 38060
    },
    {
      "epoch": 15.206152616859768,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 38061
    },
    {
      "epoch": 15.206552137435079,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 38062
    },
    {
      "epoch": 15.206951658010388,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 38063
    },
    {
      "epoch": 15.207351178585697,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 38064
    },
    {
      "epoch": 15.207750699161007,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 38065
    },
    {
      "epoch": 15.208150219736316,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 38066
    },
    {
      "epoch": 15.208549740311627,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 38067
    },
    {
      "epoch": 15.208949260886936,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 38068
    },
    {
      "epoch": 15.209348781462245,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 38069
    },
    {
      "epoch": 15.209748302037555,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 38070
    },
    {
      "epoch": 15.210147822612864,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 38071
    },
    {
      "epoch": 15.210547343188175,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 38072
    },
    {
      "epoch": 15.210946863763484,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 38073
    },
    {
      "epoch": 15.211346384338793,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 38074
    },
    {
      "epoch": 15.211745904914103,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 38075
    },
    {
      "epoch": 15.212145425489412,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 38076
    },
    {
      "epoch": 15.212544946064723,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 38077
    },
    {
      "epoch": 15.212944466640032,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 38078
    },
    {
      "epoch": 15.213343987215342,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 38079
    },
    {
      "epoch": 15.213743507790651,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 38080
    },
    {
      "epoch": 15.21414302836596,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 38081
    },
    {
      "epoch": 15.214542548941271,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 38082
    },
    {
      "epoch": 15.21494206951658,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 38083
    },
    {
      "epoch": 15.21534159009189,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 38084
    },
    {
      "epoch": 15.2157411106672,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 38085
    },
    {
      "epoch": 15.216140631242508,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 38086
    },
    {
      "epoch": 15.216540151817819,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 38087
    },
    {
      "epoch": 15.216939672393128,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 38088
    },
    {
      "epoch": 15.217339192968439,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 38089
    },
    {
      "epoch": 15.217738713543747,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 38090
    },
    {
      "epoch": 15.218138234119056,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 38091
    },
    {
      "epoch": 15.218537754694367,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 38092
    },
    {
      "epoch": 15.218937275269676,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 38093
    },
    {
      "epoch": 15.219336795844987,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 38094
    },
    {
      "epoch": 15.219736316420295,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 38095
    },
    {
      "epoch": 15.220135836995606,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 38096
    },
    {
      "epoch": 15.220535357570915,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 38097
    },
    {
      "epoch": 15.220934878146224,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 38098
    },
    {
      "epoch": 15.221334398721535,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 38099
    },
    {
      "epoch": 15.221733919296843,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 38100
    },
    {
      "epoch": 15.222133439872154,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 38101
    },
    {
      "epoch": 15.222532960447463,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 38102
    },
    {
      "epoch": 15.222932481022772,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 38103
    },
    {
      "epoch": 15.223332001598083,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 38104
    },
    {
      "epoch": 15.223731522173392,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 38105
    },
    {
      "epoch": 15.224131042748702,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 38106
    },
    {
      "epoch": 15.224530563324011,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 38107
    },
    {
      "epoch": 15.22493008389932,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 38108
    },
    {
      "epoch": 15.22532960447463,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 38109
    },
    {
      "epoch": 15.22572912504994,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 38110
    },
    {
      "epoch": 15.22612864562525,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 38111
    },
    {
      "epoch": 15.22652816620056,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 38112
    },
    {
      "epoch": 15.22692768677587,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 38113
    },
    {
      "epoch": 15.227327207351179,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 38114
    },
    {
      "epoch": 15.227726727926488,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 38115
    },
    {
      "epoch": 15.228126248501798,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 38116
    },
    {
      "epoch": 15.228525769077107,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 38117
    },
    {
      "epoch": 15.228925289652418,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 38118
    },
    {
      "epoch": 15.229324810227727,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 38119
    },
    {
      "epoch": 15.229724330803036,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 38120
    },
    {
      "epoch": 15.230123851378346,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 38121
    },
    {
      "epoch": 15.230523371953655,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 38122
    },
    {
      "epoch": 15.230922892528966,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 38123
    },
    {
      "epoch": 15.231322413104275,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 38124
    },
    {
      "epoch": 15.231721933679584,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 38125
    },
    {
      "epoch": 15.232121454254894,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 38126
    },
    {
      "epoch": 15.232520974830203,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 38127
    },
    {
      "epoch": 15.232920495405514,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 38128
    },
    {
      "epoch": 15.233320015980823,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 38129
    },
    {
      "epoch": 15.233719536556134,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 38130
    },
    {
      "epoch": 15.234119057131442,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 38131
    },
    {
      "epoch": 15.234518577706751,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 38132
    },
    {
      "epoch": 15.234918098282062,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 38133
    },
    {
      "epoch": 15.235317618857371,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 38134
    },
    {
      "epoch": 15.235717139432682,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 38135
    },
    {
      "epoch": 15.23611666000799,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 38136
    },
    {
      "epoch": 15.2365161805833,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 38137
    },
    {
      "epoch": 15.23691570115861,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 38138
    },
    {
      "epoch": 15.237315221733919,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 38139
    },
    {
      "epoch": 15.23771474230923,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 38140
    },
    {
      "epoch": 15.238114262884539,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 38141
    },
    {
      "epoch": 15.238513783459847,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 38142
    },
    {
      "epoch": 15.238913304035158,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 38143
    },
    {
      "epoch": 15.239312824610467,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 38144
    },
    {
      "epoch": 15.239712345185778,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 38145
    },
    {
      "epoch": 15.240111865761087,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 38146
    },
    {
      "epoch": 15.240511386336395,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 38147
    },
    {
      "epoch": 15.240910906911706,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 38148
    },
    {
      "epoch": 15.241310427487015,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 38149
    },
    {
      "epoch": 15.241709948062326,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 38150
    },
    {
      "epoch": 15.242109468637635,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 38151
    },
    {
      "epoch": 15.242508989212945,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 38152
    },
    {
      "epoch": 15.242908509788254,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 38153
    },
    {
      "epoch": 15.243308030363563,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 38154
    },
    {
      "epoch": 15.243707550938874,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 38155
    },
    {
      "epoch": 15.244107071514183,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 38156
    },
    {
      "epoch": 15.244506592089493,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 38157
    },
    {
      "epoch": 15.244906112664802,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 38158
    },
    {
      "epoch": 15.245305633240111,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 38159
    },
    {
      "epoch": 15.245705153815422,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 38160
    },
    {
      "epoch": 15.24610467439073,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 38161
    },
    {
      "epoch": 15.246504194966041,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 38162
    },
    {
      "epoch": 15.24690371554135,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 38163
    },
    {
      "epoch": 15.24730323611666,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 38164
    },
    {
      "epoch": 15.24770275669197,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 38165
    },
    {
      "epoch": 15.248102277267279,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 38166
    },
    {
      "epoch": 15.24850179784259,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 38167
    },
    {
      "epoch": 15.248901318417898,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 38168
    },
    {
      "epoch": 15.249300838993209,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 38169
    },
    {
      "epoch": 15.249700359568518,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 38170
    },
    {
      "epoch": 15.250099880143827,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 38171
    },
    {
      "epoch": 15.250499400719137,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 38172
    },
    {
      "epoch": 15.250898921294446,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 38173
    },
    {
      "epoch": 15.251298441869757,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 38174
    },
    {
      "epoch": 15.251697962445066,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 38175
    },
    {
      "epoch": 15.252097483020375,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 38176
    },
    {
      "epoch": 15.252497003595685,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 38177
    },
    {
      "epoch": 15.252896524170994,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 38178
    },
    {
      "epoch": 15.253296044746305,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 38179
    },
    {
      "epoch": 15.253695565321614,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 38180
    },
    {
      "epoch": 15.254095085896923,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 38181
    },
    {
      "epoch": 15.254494606472234,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 38182
    },
    {
      "epoch": 15.254894127047542,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 38183
    },
    {
      "epoch": 15.255293647622853,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 38184
    },
    {
      "epoch": 15.255693168198162,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 38185
    },
    {
      "epoch": 15.256092688773473,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0113,
      "step": 38186
    },
    {
      "epoch": 15.256492209348782,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 38187
    },
    {
      "epoch": 15.25689172992409,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 38188
    },
    {
      "epoch": 15.257291250499401,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 38189
    },
    {
      "epoch": 15.25769077107471,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 38190
    },
    {
      "epoch": 15.25809029165002,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 38191
    },
    {
      "epoch": 15.25848981222533,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 38192
    },
    {
      "epoch": 15.258889332800639,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 38193
    },
    {
      "epoch": 15.25928885337595,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 38194
    },
    {
      "epoch": 15.259688373951258,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 38195
    },
    {
      "epoch": 15.260087894526569,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 38196
    },
    {
      "epoch": 15.260487415101878,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 38197
    },
    {
      "epoch": 15.260886935677187,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 38198
    },
    {
      "epoch": 15.261286456252497,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 38199
    },
    {
      "epoch": 15.261685976827806,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 38200
    },
    {
      "epoch": 15.262085497403117,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 38201
    },
    {
      "epoch": 15.262485017978426,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 38202
    },
    {
      "epoch": 15.262884538553735,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 38203
    },
    {
      "epoch": 15.263284059129045,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 38204
    },
    {
      "epoch": 15.263683579704354,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 38205
    },
    {
      "epoch": 15.264083100279665,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 38206
    },
    {
      "epoch": 15.264482620854974,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 38207
    },
    {
      "epoch": 15.264882141430284,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0113,
      "step": 38208
    },
    {
      "epoch": 15.265281662005593,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 38209
    },
    {
      "epoch": 15.265681182580902,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 38210
    },
    {
      "epoch": 15.266080703156213,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 38211
    },
    {
      "epoch": 15.266480223731522,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 38212
    },
    {
      "epoch": 15.266879744306832,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 38213
    },
    {
      "epoch": 15.267279264882141,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 38214
    },
    {
      "epoch": 15.26767878545745,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 38215
    },
    {
      "epoch": 15.268078306032761,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 38216
    },
    {
      "epoch": 15.26847782660807,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 38217
    },
    {
      "epoch": 15.26887734718338,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 38218
    },
    {
      "epoch": 15.26927686775869,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 38219
    },
    {
      "epoch": 15.269676388333998,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 38220
    },
    {
      "epoch": 15.270075908909309,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 38221
    },
    {
      "epoch": 15.270475429484618,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 38222
    },
    {
      "epoch": 15.270874950059929,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 38223
    },
    {
      "epoch": 15.271274470635237,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 38224
    },
    {
      "epoch": 15.271673991210548,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 38225
    },
    {
      "epoch": 15.272073511785857,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 38226
    },
    {
      "epoch": 15.272473032361166,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 38227
    },
    {
      "epoch": 15.272872552936477,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 38228
    },
    {
      "epoch": 15.273272073511786,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 38229
    },
    {
      "epoch": 15.273671594087096,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 38230
    },
    {
      "epoch": 15.274071114662405,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 38231
    },
    {
      "epoch": 15.274470635237714,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 38232
    },
    {
      "epoch": 15.274870155813025,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 38233
    },
    {
      "epoch": 15.275269676388334,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 38234
    },
    {
      "epoch": 15.275669196963644,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 38235
    },
    {
      "epoch": 15.276068717538953,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 38236
    },
    {
      "epoch": 15.276468238114262,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 38237
    },
    {
      "epoch": 15.276867758689573,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 38238
    },
    {
      "epoch": 15.277267279264882,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 38239
    },
    {
      "epoch": 15.277666799840192,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 38240
    },
    {
      "epoch": 15.278066320415501,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 38241
    },
    {
      "epoch": 15.278465840990812,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 38242
    },
    {
      "epoch": 15.27886536156612,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 38243
    },
    {
      "epoch": 15.27926488214143,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 38244
    },
    {
      "epoch": 15.27966440271674,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 38245
    },
    {
      "epoch": 15.28006392329205,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 38246
    },
    {
      "epoch": 15.28046344386736,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 38247
    },
    {
      "epoch": 15.280862964442669,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 38248
    },
    {
      "epoch": 15.281262485017978,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 38249
    },
    {
      "epoch": 15.281662005593288,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 38250
    },
    {
      "epoch": 15.282061526168597,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 38251
    },
    {
      "epoch": 15.282461046743908,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 38252
    },
    {
      "epoch": 15.282860567319217,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 38253
    },
    {
      "epoch": 15.283260087894526,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 38254
    },
    {
      "epoch": 15.283659608469836,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 38255
    },
    {
      "epoch": 15.284059129045145,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 38256
    },
    {
      "epoch": 15.284458649620456,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 38257
    },
    {
      "epoch": 15.284858170195765,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 38258
    },
    {
      "epoch": 15.285257690771076,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 38259
    },
    {
      "epoch": 15.285657211346384,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 38260
    },
    {
      "epoch": 15.286056731921693,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 38261
    },
    {
      "epoch": 15.286456252497004,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 38262
    },
    {
      "epoch": 15.286855773072313,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 38263
    },
    {
      "epoch": 15.287255293647624,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 38264
    },
    {
      "epoch": 15.287654814222932,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 38265
    },
    {
      "epoch": 15.288054334798241,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0117,
      "step": 38266
    },
    {
      "epoch": 15.288453855373552,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 38267
    },
    {
      "epoch": 15.288853375948861,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 38268
    },
    {
      "epoch": 15.289252896524172,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 38269
    },
    {
      "epoch": 15.28965241709948,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 38270
    },
    {
      "epoch": 15.29005193767479,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0121,
      "step": 38271
    },
    {
      "epoch": 15.2904514582501,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 38272
    },
    {
      "epoch": 15.290850978825409,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 38273
    },
    {
      "epoch": 15.29125049940072,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 38274
    },
    {
      "epoch": 15.291650019976029,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 38275
    },
    {
      "epoch": 15.29204954055134,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 38276
    },
    {
      "epoch": 15.292449061126648,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 38277
    },
    {
      "epoch": 15.292848581701957,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 38278
    },
    {
      "epoch": 15.293248102277268,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 38279
    },
    {
      "epoch": 15.293647622852577,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 38280
    },
    {
      "epoch": 15.294047143427887,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 38281
    },
    {
      "epoch": 15.294446664003196,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 38282
    },
    {
      "epoch": 15.294846184578505,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 38283
    },
    {
      "epoch": 15.295245705153816,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 38284
    },
    {
      "epoch": 15.295645225729125,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 38285
    },
    {
      "epoch": 15.296044746304435,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 38286
    },
    {
      "epoch": 15.296444266879744,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 38287
    },
    {
      "epoch": 15.296843787455053,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 38288
    },
    {
      "epoch": 15.297243308030364,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 38289
    },
    {
      "epoch": 15.297642828605673,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 38290
    },
    {
      "epoch": 15.298042349180983,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 38291
    },
    {
      "epoch": 15.298441869756292,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 38292
    },
    {
      "epoch": 15.298841390331603,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 38293
    },
    {
      "epoch": 15.299240910906912,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 38294
    },
    {
      "epoch": 15.29964043148222,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 38295
    },
    {
      "epoch": 15.300039952057531,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 38296
    },
    {
      "epoch": 15.30043947263284,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 38297
    },
    {
      "epoch": 15.300838993208151,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 38298
    },
    {
      "epoch": 15.30123851378346,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 38299
    },
    {
      "epoch": 15.301638034358769,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 38300
    },
    {
      "epoch": 15.30203755493408,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 38301
    },
    {
      "epoch": 15.302437075509388,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 38302
    },
    {
      "epoch": 15.302836596084699,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 38303
    },
    {
      "epoch": 15.303236116660008,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 38304
    },
    {
      "epoch": 15.303635637235317,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 38305
    },
    {
      "epoch": 15.304035157810628,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 38306
    },
    {
      "epoch": 15.304434678385936,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 38307
    },
    {
      "epoch": 15.304834198961247,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 38308
    },
    {
      "epoch": 15.305233719536556,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 38309
    },
    {
      "epoch": 15.305633240111867,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 38310
    },
    {
      "epoch": 15.306032760687176,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 38311
    },
    {
      "epoch": 15.306432281262484,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 38312
    },
    {
      "epoch": 15.306831801837795,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 38313
    },
    {
      "epoch": 15.307231322413104,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 38314
    },
    {
      "epoch": 15.307630842988415,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 38315
    },
    {
      "epoch": 15.308030363563724,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 38316
    },
    {
      "epoch": 15.308429884139032,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 38317
    },
    {
      "epoch": 15.308829404714343,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 38318
    },
    {
      "epoch": 15.309228925289652,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 38319
    },
    {
      "epoch": 15.309628445864963,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 38320
    },
    {
      "epoch": 15.310027966440272,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 38321
    },
    {
      "epoch": 15.31042748701558,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 38322
    },
    {
      "epoch": 15.310827007590891,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 38323
    },
    {
      "epoch": 15.3112265281662,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 38324
    },
    {
      "epoch": 15.31162604874151,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 38325
    },
    {
      "epoch": 15.31202556931682,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 38326
    },
    {
      "epoch": 15.31242508989213,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 38327
    },
    {
      "epoch": 15.31282461046744,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 38328
    },
    {
      "epoch": 15.313224131042748,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 38329
    },
    {
      "epoch": 15.313623651618059,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 38330
    },
    {
      "epoch": 15.314023172193368,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 38331
    },
    {
      "epoch": 15.314422692768678,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.01,
      "step": 38332
    },
    {
      "epoch": 15.314822213343987,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 38333
    },
    {
      "epoch": 15.315221733919296,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 38334
    },
    {
      "epoch": 15.315621254494607,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 38335
    },
    {
      "epoch": 15.316020775069916,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 38336
    },
    {
      "epoch": 15.316420295645226,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 38337
    },
    {
      "epoch": 15.316819816220535,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 38338
    },
    {
      "epoch": 15.317219336795844,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 38339
    },
    {
      "epoch": 15.317618857371155,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 38340
    },
    {
      "epoch": 15.318018377946464,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 38341
    },
    {
      "epoch": 15.318417898521774,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0127,
      "step": 38342
    },
    {
      "epoch": 15.318817419097083,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 38343
    },
    {
      "epoch": 15.319216939672392,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 38344
    },
    {
      "epoch": 15.319616460247703,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 38345
    },
    {
      "epoch": 15.320015980823012,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 38346
    },
    {
      "epoch": 15.320415501398323,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 38347
    },
    {
      "epoch": 15.320815021973631,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 38348
    },
    {
      "epoch": 15.321214542548942,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 38349
    },
    {
      "epoch": 15.321614063124251,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 38350
    },
    {
      "epoch": 15.32201358369956,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 38351
    },
    {
      "epoch": 15.32241310427487,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 38352
    },
    {
      "epoch": 15.32281262485018,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 38353
    },
    {
      "epoch": 15.32321214542549,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 38354
    },
    {
      "epoch": 15.323611666000799,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 38355
    },
    {
      "epoch": 15.324011186576108,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 38356
    },
    {
      "epoch": 15.324410707151419,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0111,
      "step": 38357
    },
    {
      "epoch": 15.324810227726728,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 38358
    },
    {
      "epoch": 15.325209748302038,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 38359
    },
    {
      "epoch": 15.325609268877347,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 38360
    },
    {
      "epoch": 15.326008789452656,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 38361
    },
    {
      "epoch": 15.326408310027967,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 38362
    },
    {
      "epoch": 15.326807830603276,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 38363
    },
    {
      "epoch": 15.327207351178586,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 38364
    },
    {
      "epoch": 15.327606871753895,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 38365
    },
    {
      "epoch": 15.328006392329206,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 38366
    },
    {
      "epoch": 15.328405912904515,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0083,
      "step": 38367
    },
    {
      "epoch": 15.328805433479824,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 38368
    },
    {
      "epoch": 15.329204954055134,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 38369
    },
    {
      "epoch": 15.329604474630443,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 38370
    },
    {
      "epoch": 15.330003995205754,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 38371
    },
    {
      "epoch": 15.330403515781063,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 38372
    },
    {
      "epoch": 15.330803036356372,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 38373
    },
    {
      "epoch": 15.331202556931682,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 38374
    },
    {
      "epoch": 15.331602077506991,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 38375
    },
    {
      "epoch": 15.332001598082302,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 38376
    },
    {
      "epoch": 15.33240111865761,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 38377
    },
    {
      "epoch": 15.33280063923292,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 38378
    },
    {
      "epoch": 15.33320015980823,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 38379
    },
    {
      "epoch": 15.33359968038354,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 38380
    },
    {
      "epoch": 15.33399920095885,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 38381
    },
    {
      "epoch": 15.334398721534159,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 38382
    },
    {
      "epoch": 15.334798242109468,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 38383
    },
    {
      "epoch": 15.335197762684778,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 38384
    },
    {
      "epoch": 15.335597283260087,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 38385
    },
    {
      "epoch": 15.335996803835398,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 38386
    },
    {
      "epoch": 15.336396324410707,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 38387
    },
    {
      "epoch": 15.336795844986018,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 38388
    },
    {
      "epoch": 15.337195365561326,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 38389
    },
    {
      "epoch": 15.337594886136635,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 38390
    },
    {
      "epoch": 15.337994406711946,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 38391
    },
    {
      "epoch": 15.338393927287255,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 38392
    },
    {
      "epoch": 15.338793447862566,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 38393
    },
    {
      "epoch": 15.339192968437874,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 38394
    },
    {
      "epoch": 15.339592489013183,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 38395
    },
    {
      "epoch": 15.339992009588494,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 38396
    },
    {
      "epoch": 15.340391530163803,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 38397
    },
    {
      "epoch": 15.340791050739114,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 38398
    },
    {
      "epoch": 15.341190571314423,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 38399
    },
    {
      "epoch": 15.341590091889731,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 38400
    },
    {
      "epoch": 15.341989612465042,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 38401
    },
    {
      "epoch": 15.342389133040351,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 38402
    },
    {
      "epoch": 15.342788653615662,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 38403
    },
    {
      "epoch": 15.34318817419097,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 38404
    },
    {
      "epoch": 15.343587694766281,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 38405
    },
    {
      "epoch": 15.34398721534159,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 38406
    },
    {
      "epoch": 15.344386735916899,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 38407
    },
    {
      "epoch": 15.34478625649221,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 38408
    },
    {
      "epoch": 15.345185777067519,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 38409
    },
    {
      "epoch": 15.34558529764283,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 38410
    },
    {
      "epoch": 15.345984818218138,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 38411
    },
    {
      "epoch": 15.346384338793447,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 38412
    },
    {
      "epoch": 15.346783859368758,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 38413
    },
    {
      "epoch": 15.347183379944067,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 38414
    },
    {
      "epoch": 15.347582900519377,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 38415
    },
    {
      "epoch": 15.347982421094686,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 38416
    },
    {
      "epoch": 15.348381941669995,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 38417
    },
    {
      "epoch": 15.348781462245306,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 38418
    },
    {
      "epoch": 15.349180982820615,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 38419
    },
    {
      "epoch": 15.349580503395925,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 38420
    },
    {
      "epoch": 15.349980023971234,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 38421
    },
    {
      "epoch": 15.350379544546545,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 38422
    },
    {
      "epoch": 15.350779065121854,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 38423
    },
    {
      "epoch": 15.351178585697163,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 38424
    },
    {
      "epoch": 15.351578106272473,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 38425
    },
    {
      "epoch": 15.351977626847782,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 38426
    },
    {
      "epoch": 15.352377147423093,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 38427
    },
    {
      "epoch": 15.352776667998402,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 38428
    },
    {
      "epoch": 15.35317618857371,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 38429
    },
    {
      "epoch": 15.353575709149021,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 38430
    },
    {
      "epoch": 15.35397522972433,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 38431
    },
    {
      "epoch": 15.354374750299641,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 38432
    },
    {
      "epoch": 15.35477427087495,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 38433
    },
    {
      "epoch": 15.355173791450259,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 38434
    },
    {
      "epoch": 15.35557331202557,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 38435
    },
    {
      "epoch": 15.355972832600878,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 38436
    },
    {
      "epoch": 15.356372353176189,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 38437
    },
    {
      "epoch": 15.356771873751498,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 38438
    },
    {
      "epoch": 15.357171394326809,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 38439
    },
    {
      "epoch": 15.357570914902118,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 38440
    },
    {
      "epoch": 15.357970435477426,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 38441
    },
    {
      "epoch": 15.358369956052737,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 38442
    },
    {
      "epoch": 15.358769476628046,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 38443
    },
    {
      "epoch": 15.359168997203357,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 38444
    },
    {
      "epoch": 15.359568517778666,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 38445
    },
    {
      "epoch": 15.359968038353975,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 38446
    },
    {
      "epoch": 15.360367558929285,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 38447
    },
    {
      "epoch": 15.360767079504594,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 38448
    },
    {
      "epoch": 15.361166600079905,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 38449
    },
    {
      "epoch": 15.361566120655214,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 38450
    },
    {
      "epoch": 15.361965641230523,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 38451
    },
    {
      "epoch": 15.362365161805833,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 38452
    },
    {
      "epoch": 15.362764682381142,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 38453
    },
    {
      "epoch": 15.363164202956453,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 38454
    },
    {
      "epoch": 15.363563723531762,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 38455
    },
    {
      "epoch": 15.363963244107072,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 38456
    },
    {
      "epoch": 15.364362764682381,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 38457
    },
    {
      "epoch": 15.36476228525769,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 38458
    },
    {
      "epoch": 15.365161805833,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 38459
    },
    {
      "epoch": 15.36556132640831,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 38460
    },
    {
      "epoch": 15.36596084698362,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 38461
    },
    {
      "epoch": 15.36636036755893,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 38462
    },
    {
      "epoch": 15.366759888134238,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 38463
    },
    {
      "epoch": 15.367159408709549,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 38464
    },
    {
      "epoch": 15.367558929284858,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 38465
    },
    {
      "epoch": 15.367958449860168,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 38466
    },
    {
      "epoch": 15.368357970435477,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 38467
    },
    {
      "epoch": 15.368757491010786,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 38468
    },
    {
      "epoch": 15.369157011586097,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 38469
    },
    {
      "epoch": 15.369556532161406,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 38470
    },
    {
      "epoch": 15.369956052736717,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 38471
    },
    {
      "epoch": 15.370355573312025,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 38472
    },
    {
      "epoch": 15.370755093887336,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 38473
    },
    {
      "epoch": 15.371154614462645,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 38474
    },
    {
      "epoch": 15.371554135037954,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 38475
    },
    {
      "epoch": 15.371953655613265,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 38476
    },
    {
      "epoch": 15.372353176188573,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 38477
    },
    {
      "epoch": 15.372752696763884,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 38478
    },
    {
      "epoch": 15.373152217339193,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 38479
    },
    {
      "epoch": 15.373551737914502,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 38480
    },
    {
      "epoch": 15.373951258489813,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 38481
    },
    {
      "epoch": 15.374350779065121,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 38482
    },
    {
      "epoch": 15.374750299640432,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 38483
    },
    {
      "epoch": 15.375149820215741,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 38484
    },
    {
      "epoch": 15.37554934079105,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 38485
    },
    {
      "epoch": 15.37594886136636,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 38486
    },
    {
      "epoch": 15.37634838194167,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 38487
    },
    {
      "epoch": 15.37674790251698,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 38488
    },
    {
      "epoch": 15.377147423092289,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 38489
    },
    {
      "epoch": 15.3775469436676,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 38490
    },
    {
      "epoch": 15.377946464242909,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 38491
    },
    {
      "epoch": 15.378345984818218,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 38492
    },
    {
      "epoch": 15.378745505393528,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 38493
    },
    {
      "epoch": 15.379145025968837,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 38494
    },
    {
      "epoch": 15.379544546544148,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 38495
    },
    {
      "epoch": 15.379944067119457,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 38496
    },
    {
      "epoch": 15.380343587694766,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 38497
    },
    {
      "epoch": 15.380743108270076,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 38498
    },
    {
      "epoch": 15.381142628845385,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 38499
    },
    {
      "epoch": 15.381542149420696,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 38500
    },
    {
      "epoch": 15.381941669996005,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 38501
    },
    {
      "epoch": 15.382341190571314,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 38502
    },
    {
      "epoch": 15.382740711146624,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 38503
    },
    {
      "epoch": 15.383140231721933,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 38504
    },
    {
      "epoch": 15.383539752297244,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 38505
    },
    {
      "epoch": 15.383939272872553,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 38506
    },
    {
      "epoch": 15.384338793447863,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 38507
    },
    {
      "epoch": 15.384738314023172,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 38508
    },
    {
      "epoch": 15.385137834598481,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 38509
    },
    {
      "epoch": 15.385537355173792,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 38510
    },
    {
      "epoch": 15.3859368757491,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 38511
    },
    {
      "epoch": 15.386336396324412,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 38512
    },
    {
      "epoch": 15.38673591689972,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 38513
    },
    {
      "epoch": 15.38713543747503,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 38514
    },
    {
      "epoch": 15.38753495805034,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 38515
    },
    {
      "epoch": 15.387934478625649,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 38516
    },
    {
      "epoch": 15.38833399920096,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 38517
    },
    {
      "epoch": 15.388733519776268,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 38518
    },
    {
      "epoch": 15.389133040351577,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 38519
    },
    {
      "epoch": 15.389532560926888,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 38520
    },
    {
      "epoch": 15.389932081502197,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 38521
    },
    {
      "epoch": 15.390331602077508,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 38522
    },
    {
      "epoch": 15.390731122652817,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 38523
    },
    {
      "epoch": 15.391130643228125,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 38524
    },
    {
      "epoch": 15.391530163803436,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 38525
    },
    {
      "epoch": 15.391929684378745,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 38526
    },
    {
      "epoch": 15.392329204954056,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 38527
    },
    {
      "epoch": 15.392728725529365,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 38528
    },
    {
      "epoch": 15.393128246104675,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 38529
    },
    {
      "epoch": 15.393527766679984,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 38530
    },
    {
      "epoch": 15.393927287255293,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 38531
    },
    {
      "epoch": 15.394326807830604,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 38532
    },
    {
      "epoch": 15.394726328405913,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 38533
    },
    {
      "epoch": 15.395125848981223,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 38534
    },
    {
      "epoch": 15.395525369556532,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 38535
    },
    {
      "epoch": 15.395924890131841,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 38536
    },
    {
      "epoch": 15.396324410707152,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 38537
    },
    {
      "epoch": 15.39672393128246,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 38538
    },
    {
      "epoch": 15.397123451857771,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 38539
    },
    {
      "epoch": 15.39752297243308,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 38540
    },
    {
      "epoch": 15.397922493008389,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 38541
    },
    {
      "epoch": 15.3983220135837,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 38542
    },
    {
      "epoch": 15.398721534159009,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 38543
    },
    {
      "epoch": 15.39912105473432,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 38544
    },
    {
      "epoch": 15.399520575309628,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 38545
    },
    {
      "epoch": 15.399920095884939,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 38546
    },
    {
      "epoch": 15.400319616460248,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 38547
    },
    {
      "epoch": 15.400719137035557,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 38548
    },
    {
      "epoch": 15.401118657610867,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 38549
    },
    {
      "epoch": 15.401518178186176,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 38550
    },
    {
      "epoch": 15.401917698761487,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 38551
    },
    {
      "epoch": 15.402317219336796,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 38552
    },
    {
      "epoch": 15.402716739912105,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 38553
    },
    {
      "epoch": 15.403116260487415,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 38554
    },
    {
      "epoch": 15.403515781062724,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 38555
    },
    {
      "epoch": 15.403915301638035,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 38556
    },
    {
      "epoch": 15.404314822213344,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 38557
    },
    {
      "epoch": 15.404714342788653,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 38558
    },
    {
      "epoch": 15.405113863363963,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 38559
    },
    {
      "epoch": 15.405513383939272,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 38560
    },
    {
      "epoch": 15.405912904514583,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 38561
    },
    {
      "epoch": 15.406312425089892,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 38562
    },
    {
      "epoch": 15.4067119456652,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 38563
    },
    {
      "epoch": 15.407111466240512,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 38564
    },
    {
      "epoch": 15.40751098681582,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 38565
    },
    {
      "epoch": 15.407910507391131,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 38566
    },
    {
      "epoch": 15.40831002796644,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 38567
    },
    {
      "epoch": 15.40870954854175,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 38568
    },
    {
      "epoch": 15.40910906911706,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 38569
    },
    {
      "epoch": 15.409508589692368,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 38570
    },
    {
      "epoch": 15.40990811026768,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 38571
    },
    {
      "epoch": 15.410307630842988,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 38572
    },
    {
      "epoch": 15.410707151418299,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 38573
    },
    {
      "epoch": 15.411106671993608,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 38574
    },
    {
      "epoch": 15.411506192568917,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 38575
    },
    {
      "epoch": 15.411905713144227,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 38576
    },
    {
      "epoch": 15.412305233719536,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 38577
    },
    {
      "epoch": 15.412704754294847,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 38578
    },
    {
      "epoch": 15.413104274870156,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 38579
    },
    {
      "epoch": 15.413503795445465,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 38580
    },
    {
      "epoch": 15.413903316020775,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 38581
    },
    {
      "epoch": 15.414302836596084,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 38582
    },
    {
      "epoch": 15.414702357171395,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 38583
    },
    {
      "epoch": 15.415101877746704,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 38584
    },
    {
      "epoch": 15.415501398322014,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 38585
    },
    {
      "epoch": 15.415900918897323,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 38586
    },
    {
      "epoch": 15.416300439472632,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 38587
    },
    {
      "epoch": 15.416699960047943,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 38588
    },
    {
      "epoch": 15.417099480623252,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 38589
    },
    {
      "epoch": 15.417499001198562,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 38590
    },
    {
      "epoch": 15.417898521773871,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 38591
    },
    {
      "epoch": 15.41829804234918,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 38592
    },
    {
      "epoch": 15.418697562924491,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 38593
    },
    {
      "epoch": 15.4190970834998,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 38594
    },
    {
      "epoch": 15.41949660407511,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 38595
    },
    {
      "epoch": 15.41989612465042,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 38596
    },
    {
      "epoch": 15.420295645225728,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 38597
    },
    {
      "epoch": 15.420695165801039,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 38598
    },
    {
      "epoch": 15.421094686376348,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 38599
    },
    {
      "epoch": 15.421494206951659,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 38600
    },
    {
      "epoch": 15.421893727526967,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 38601
    },
    {
      "epoch": 15.422293248102278,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 38602
    },
    {
      "epoch": 15.422692768677587,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 38603
    },
    {
      "epoch": 15.423092289252896,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 38604
    },
    {
      "epoch": 15.423491809828207,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 38605
    },
    {
      "epoch": 15.423891330403515,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 38606
    },
    {
      "epoch": 15.424290850978826,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 38607
    },
    {
      "epoch": 15.424690371554135,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 38608
    },
    {
      "epoch": 15.425089892129444,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 38609
    },
    {
      "epoch": 15.425489412704755,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 38610
    },
    {
      "epoch": 15.425888933280063,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 38611
    },
    {
      "epoch": 15.426288453855374,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 38612
    },
    {
      "epoch": 15.426687974430683,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 38613
    },
    {
      "epoch": 15.427087495005992,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 38614
    },
    {
      "epoch": 15.427487015581303,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 38615
    },
    {
      "epoch": 15.427886536156612,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0088,
      "step": 38616
    },
    {
      "epoch": 15.428286056731922,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 38617
    },
    {
      "epoch": 15.428685577307231,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 38618
    },
    {
      "epoch": 15.429085097882542,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 38619
    },
    {
      "epoch": 15.42948461845785,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 38620
    },
    {
      "epoch": 15.42988413903316,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 38621
    },
    {
      "epoch": 15.43028365960847,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 38622
    },
    {
      "epoch": 15.43068318018378,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 38623
    },
    {
      "epoch": 15.43108270075909,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 38624
    },
    {
      "epoch": 15.431482221334399,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 38625
    },
    {
      "epoch": 15.431881741909708,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 38626
    },
    {
      "epoch": 15.432281262485018,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 38627
    },
    {
      "epoch": 15.432680783060327,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 38628
    },
    {
      "epoch": 15.433080303635638,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 38629
    },
    {
      "epoch": 15.433479824210947,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 38630
    },
    {
      "epoch": 15.433879344786256,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 38631
    },
    {
      "epoch": 15.434278865361566,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 38632
    },
    {
      "epoch": 15.434678385936875,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 38633
    },
    {
      "epoch": 15.435077906512186,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 38634
    },
    {
      "epoch": 15.435477427087495,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 38635
    },
    {
      "epoch": 15.435876947662805,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 38636
    },
    {
      "epoch": 15.436276468238114,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 38637
    },
    {
      "epoch": 15.436675988813423,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 38638
    },
    {
      "epoch": 15.437075509388734,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 38639
    },
    {
      "epoch": 15.437475029964043,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 38640
    },
    {
      "epoch": 15.437874550539354,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 38641
    },
    {
      "epoch": 15.438274071114662,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 38642
    },
    {
      "epoch": 15.438673591689971,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 38643
    },
    {
      "epoch": 15.439073112265282,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 38644
    },
    {
      "epoch": 15.439472632840591,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 38645
    },
    {
      "epoch": 15.439872153415902,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 38646
    },
    {
      "epoch": 15.44027167399121,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 38647
    },
    {
      "epoch": 15.44067119456652,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 38648
    },
    {
      "epoch": 15.44107071514183,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 38649
    },
    {
      "epoch": 15.441470235717139,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 38650
    },
    {
      "epoch": 15.44186975629245,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 38651
    },
    {
      "epoch": 15.442269276867759,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 38652
    },
    {
      "epoch": 15.44266879744307,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 38653
    },
    {
      "epoch": 15.443068318018378,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 38654
    },
    {
      "epoch": 15.443467838593687,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 38655
    },
    {
      "epoch": 15.443867359168998,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 38656
    },
    {
      "epoch": 15.444266879744307,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 38657
    },
    {
      "epoch": 15.444666400319617,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 38658
    },
    {
      "epoch": 15.445065920894926,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 38659
    },
    {
      "epoch": 15.445465441470235,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 38660
    },
    {
      "epoch": 15.445864962045546,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 38661
    },
    {
      "epoch": 15.446264482620855,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 38662
    },
    {
      "epoch": 15.446664003196165,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 38663
    },
    {
      "epoch": 15.447063523771474,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 38664
    },
    {
      "epoch": 15.447463044346783,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 38665
    },
    {
      "epoch": 15.447862564922094,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 38666
    },
    {
      "epoch": 15.448262085497403,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 38667
    },
    {
      "epoch": 15.448661606072713,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 38668
    },
    {
      "epoch": 15.449061126648022,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 38669
    },
    {
      "epoch": 15.449460647223333,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 38670
    },
    {
      "epoch": 15.449860167798642,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 38671
    },
    {
      "epoch": 15.45025968837395,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 38672
    },
    {
      "epoch": 15.450659208949261,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 38673
    },
    {
      "epoch": 15.45105872952457,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 38674
    },
    {
      "epoch": 15.451458250099881,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 38675
    },
    {
      "epoch": 15.45185777067519,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 38676
    },
    {
      "epoch": 15.452257291250499,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 38677
    },
    {
      "epoch": 15.45265681182581,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 38678
    },
    {
      "epoch": 15.453056332401118,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 38679
    },
    {
      "epoch": 15.453455852976429,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 38680
    },
    {
      "epoch": 15.453855373551738,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 38681
    },
    {
      "epoch": 15.454254894127047,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 38682
    },
    {
      "epoch": 15.454654414702357,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 38683
    },
    {
      "epoch": 15.455053935277666,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 38684
    },
    {
      "epoch": 15.455453455852977,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 38685
    },
    {
      "epoch": 15.455852976428286,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 38686
    },
    {
      "epoch": 15.456252497003595,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 38687
    },
    {
      "epoch": 15.456652017578906,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 38688
    },
    {
      "epoch": 15.457051538154214,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 38689
    },
    {
      "epoch": 15.457451058729525,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 38690
    },
    {
      "epoch": 15.457850579304834,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 38691
    },
    {
      "epoch": 15.458250099880145,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 38692
    },
    {
      "epoch": 15.458649620455454,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 38693
    },
    {
      "epoch": 15.459049141030762,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 38694
    },
    {
      "epoch": 15.459448661606073,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 38695
    },
    {
      "epoch": 15.459848182181382,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 38696
    },
    {
      "epoch": 15.460247702756693,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 38697
    },
    {
      "epoch": 15.460647223332002,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 38698
    },
    {
      "epoch": 15.46104674390731,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 38699
    },
    {
      "epoch": 15.461446264482621,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 38700
    },
    {
      "epoch": 15.46184578505793,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 38701
    },
    {
      "epoch": 15.46224530563324,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 38702
    },
    {
      "epoch": 15.46264482620855,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 38703
    },
    {
      "epoch": 15.463044346783859,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 38704
    },
    {
      "epoch": 15.46344386735917,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 38705
    },
    {
      "epoch": 15.463843387934478,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 38706
    },
    {
      "epoch": 15.464242908509789,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 38707
    },
    {
      "epoch": 15.464642429085098,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 38708
    },
    {
      "epoch": 15.465041949660408,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 38709
    },
    {
      "epoch": 15.465441470235717,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 38710
    },
    {
      "epoch": 15.465840990811026,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 38711
    },
    {
      "epoch": 15.466240511386337,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 38712
    },
    {
      "epoch": 15.466640031961646,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 38713
    },
    {
      "epoch": 15.467039552536956,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 38714
    },
    {
      "epoch": 15.467439073112265,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 38715
    },
    {
      "epoch": 15.467838593687574,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 38716
    },
    {
      "epoch": 15.468238114262885,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 38717
    },
    {
      "epoch": 15.468637634838194,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 38718
    },
    {
      "epoch": 15.469037155413504,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 38719
    },
    {
      "epoch": 15.469436675988813,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 38720
    },
    {
      "epoch": 15.469836196564122,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 38721
    },
    {
      "epoch": 15.470235717139433,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 38722
    },
    {
      "epoch": 15.470635237714742,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 38723
    },
    {
      "epoch": 15.471034758290052,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 38724
    },
    {
      "epoch": 15.471434278865361,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 38725
    },
    {
      "epoch": 15.471833799440672,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 38726
    },
    {
      "epoch": 15.472233320015981,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 38727
    },
    {
      "epoch": 15.47263284059129,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 38728
    },
    {
      "epoch": 15.4730323611666,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 38729
    },
    {
      "epoch": 15.47343188174191,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 38730
    },
    {
      "epoch": 15.47383140231722,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 38731
    },
    {
      "epoch": 15.474230922892529,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 38732
    },
    {
      "epoch": 15.474630443467838,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 38733
    },
    {
      "epoch": 15.475029964043149,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 38734
    },
    {
      "epoch": 15.475429484618457,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 38735
    },
    {
      "epoch": 15.475829005193768,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 38736
    },
    {
      "epoch": 15.476228525769077,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 38737
    },
    {
      "epoch": 15.476628046344386,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 38738
    },
    {
      "epoch": 15.477027566919697,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 38739
    },
    {
      "epoch": 15.477427087495006,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 38740
    },
    {
      "epoch": 15.477826608070316,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 38741
    },
    {
      "epoch": 15.478226128645625,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 38742
    },
    {
      "epoch": 15.478625649220934,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 38743
    },
    {
      "epoch": 15.479025169796245,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 38744
    },
    {
      "epoch": 15.479424690371554,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 38745
    },
    {
      "epoch": 15.479824210946864,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 38746
    },
    {
      "epoch": 15.480223731522173,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 38747
    },
    {
      "epoch": 15.480623252097484,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 38748
    },
    {
      "epoch": 15.481022772672793,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 38749
    },
    {
      "epoch": 15.481422293248102,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 38750
    },
    {
      "epoch": 15.481821813823412,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 38751
    },
    {
      "epoch": 15.482221334398721,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 38752
    },
    {
      "epoch": 15.482620854974032,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 38753
    },
    {
      "epoch": 15.48302037554934,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 38754
    },
    {
      "epoch": 15.48341989612465,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 38755
    },
    {
      "epoch": 15.48381941669996,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 38756
    },
    {
      "epoch": 15.48421893727527,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 38757
    },
    {
      "epoch": 15.48461845785058,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 38758
    },
    {
      "epoch": 15.485017978425889,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 38759
    },
    {
      "epoch": 15.485417499001198,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 38760
    },
    {
      "epoch": 15.485817019576508,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 38761
    },
    {
      "epoch": 15.486216540151817,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 38762
    },
    {
      "epoch": 15.486616060727128,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 38763
    },
    {
      "epoch": 15.487015581302437,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 38764
    },
    {
      "epoch": 15.487415101877748,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 38765
    },
    {
      "epoch": 15.487814622453056,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 38766
    },
    {
      "epoch": 15.488214143028365,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 38767
    },
    {
      "epoch": 15.488613663603676,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 38768
    },
    {
      "epoch": 15.489013184178985,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 38769
    },
    {
      "epoch": 15.489412704754296,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 38770
    },
    {
      "epoch": 15.489812225329604,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 38771
    },
    {
      "epoch": 15.490211745904913,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 38772
    },
    {
      "epoch": 15.490611266480224,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 38773
    },
    {
      "epoch": 15.491010787055533,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 38774
    },
    {
      "epoch": 15.491410307630844,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 38775
    },
    {
      "epoch": 15.491809828206152,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 38776
    },
    {
      "epoch": 15.492209348781461,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 38777
    },
    {
      "epoch": 15.492608869356772,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 38778
    },
    {
      "epoch": 15.493008389932081,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 38779
    },
    {
      "epoch": 15.493407910507392,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 38780
    },
    {
      "epoch": 15.4938074310827,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 38781
    },
    {
      "epoch": 15.494206951658011,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 38782
    },
    {
      "epoch": 15.49460647223332,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 38783
    },
    {
      "epoch": 15.495005992808629,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 38784
    },
    {
      "epoch": 15.49540551338394,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 38785
    },
    {
      "epoch": 15.495805033959249,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 38786
    },
    {
      "epoch": 15.49620455453456,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 38787
    },
    {
      "epoch": 15.496604075109868,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.011,
      "step": 38788
    },
    {
      "epoch": 15.497003595685177,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 38789
    },
    {
      "epoch": 15.497403116260488,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 38790
    },
    {
      "epoch": 15.497802636835797,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 38791
    },
    {
      "epoch": 15.498202157411107,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 38792
    },
    {
      "epoch": 15.498601677986416,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 38793
    },
    {
      "epoch": 15.499001198561725,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 38794
    },
    {
      "epoch": 15.499400719137036,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 38795
    },
    {
      "epoch": 15.499800239712345,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 38796
    },
    {
      "epoch": 15.500199760287655,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 38797
    },
    {
      "epoch": 15.500599280862964,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 38798
    },
    {
      "epoch": 15.500998801438275,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 38799
    },
    {
      "epoch": 15.501398322013584,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 38800
    },
    {
      "epoch": 15.501797842588893,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 38801
    },
    {
      "epoch": 15.502197363164203,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 38802
    },
    {
      "epoch": 15.502596883739512,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 38803
    },
    {
      "epoch": 15.502996404314823,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 38804
    },
    {
      "epoch": 15.503395924890132,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 38805
    },
    {
      "epoch": 15.50379544546544,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 38806
    },
    {
      "epoch": 15.504194966040751,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 38807
    },
    {
      "epoch": 15.50459448661606,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 38808
    },
    {
      "epoch": 15.504994007191371,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 38809
    },
    {
      "epoch": 15.50539352776668,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 38810
    },
    {
      "epoch": 15.505793048341989,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 38811
    },
    {
      "epoch": 15.5061925689173,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 38812
    },
    {
      "epoch": 15.506592089492608,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 38813
    },
    {
      "epoch": 15.506991610067919,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 38814
    },
    {
      "epoch": 15.507391130643228,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 38815
    },
    {
      "epoch": 15.507790651218539,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 38816
    },
    {
      "epoch": 15.508190171793848,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 38817
    },
    {
      "epoch": 15.508589692369156,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 38818
    },
    {
      "epoch": 15.508989212944467,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 38819
    },
    {
      "epoch": 15.509388733519776,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 38820
    },
    {
      "epoch": 15.509788254095087,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 38821
    },
    {
      "epoch": 15.510187774670396,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 38822
    },
    {
      "epoch": 15.510587295245704,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 38823
    },
    {
      "epoch": 15.510986815821015,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 38824
    },
    {
      "epoch": 15.511386336396324,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 38825
    },
    {
      "epoch": 15.511785856971635,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 38826
    },
    {
      "epoch": 15.512185377546944,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 38827
    },
    {
      "epoch": 15.512584898122252,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 38828
    },
    {
      "epoch": 15.512984418697563,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 38829
    },
    {
      "epoch": 15.513383939272872,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 38830
    },
    {
      "epoch": 15.513783459848183,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 38831
    },
    {
      "epoch": 15.514182980423492,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 38832
    },
    {
      "epoch": 15.514582500998802,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 38833
    },
    {
      "epoch": 15.514982021574111,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 38834
    },
    {
      "epoch": 15.51538154214942,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 38835
    },
    {
      "epoch": 15.51578106272473,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 38836
    },
    {
      "epoch": 15.51618058330004,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 38837
    },
    {
      "epoch": 15.51658010387535,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 38838
    },
    {
      "epoch": 15.51697962445066,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 38839
    },
    {
      "epoch": 15.517379145025968,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 38840
    },
    {
      "epoch": 15.517778665601279,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 38841
    },
    {
      "epoch": 15.518178186176588,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 38842
    },
    {
      "epoch": 15.518577706751898,
      "grad_norm": 0.036865234375,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 38843
    },
    {
      "epoch": 15.518977227327207,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 38844
    },
    {
      "epoch": 15.519376747902516,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 38845
    },
    {
      "epoch": 15.519776268477827,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 38846
    },
    {
      "epoch": 15.520175789053136,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 38847
    },
    {
      "epoch": 15.520575309628446,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 38848
    },
    {
      "epoch": 15.520974830203755,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 38849
    },
    {
      "epoch": 15.521374350779066,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 38850
    },
    {
      "epoch": 15.521773871354375,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 38851
    },
    {
      "epoch": 15.522173391929684,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 38852
    },
    {
      "epoch": 15.522572912504994,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 38853
    },
    {
      "epoch": 15.522972433080303,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 38854
    },
    {
      "epoch": 15.523371953655614,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 38855
    },
    {
      "epoch": 15.523771474230923,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 38856
    },
    {
      "epoch": 15.524170994806232,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 38857
    },
    {
      "epoch": 15.524570515381543,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 38858
    },
    {
      "epoch": 15.524970035956851,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 38859
    },
    {
      "epoch": 15.525369556532162,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 38860
    },
    {
      "epoch": 15.525769077107471,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 38861
    },
    {
      "epoch": 15.52616859768278,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 38862
    },
    {
      "epoch": 15.52656811825809,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 38863
    },
    {
      "epoch": 15.5269676388334,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 38864
    },
    {
      "epoch": 15.52736715940871,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 38865
    },
    {
      "epoch": 15.527766679984019,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 38866
    },
    {
      "epoch": 15.52816620055933,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 38867
    },
    {
      "epoch": 15.528565721134639,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 38868
    },
    {
      "epoch": 15.528965241709948,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 38869
    },
    {
      "epoch": 15.529364762285258,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 38870
    },
    {
      "epoch": 15.529764282860567,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 38871
    },
    {
      "epoch": 15.530163803435878,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 38872
    },
    {
      "epoch": 15.530563324011187,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 38873
    },
    {
      "epoch": 15.530962844586496,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 38874
    },
    {
      "epoch": 15.531362365161806,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 38875
    },
    {
      "epoch": 15.531761885737115,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 38876
    },
    {
      "epoch": 15.532161406312426,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 38877
    },
    {
      "epoch": 15.532560926887735,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 38878
    },
    {
      "epoch": 15.532960447463044,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 38879
    },
    {
      "epoch": 15.533359968038354,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 38880
    },
    {
      "epoch": 15.533759488613663,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 38881
    },
    {
      "epoch": 15.534159009188974,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 38882
    },
    {
      "epoch": 15.534558529764283,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 38883
    },
    {
      "epoch": 15.534958050339593,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 38884
    },
    {
      "epoch": 15.535357570914902,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 38885
    },
    {
      "epoch": 15.535757091490211,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 38886
    },
    {
      "epoch": 15.536156612065522,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 38887
    },
    {
      "epoch": 15.53655613264083,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 38888
    },
    {
      "epoch": 15.53695565321614,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 38889
    },
    {
      "epoch": 15.53735517379145,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 38890
    },
    {
      "epoch": 15.53775469436676,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 38891
    },
    {
      "epoch": 15.53815421494207,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 38892
    },
    {
      "epoch": 15.538553735517379,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 38893
    },
    {
      "epoch": 15.53895325609269,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 38894
    },
    {
      "epoch": 15.539352776667998,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 38895
    },
    {
      "epoch": 15.539752297243307,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 38896
    },
    {
      "epoch": 15.540151817818618,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 38897
    },
    {
      "epoch": 15.540551338393927,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 38898
    },
    {
      "epoch": 15.540950858969238,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 38899
    },
    {
      "epoch": 15.541350379544546,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 38900
    },
    {
      "epoch": 15.541749900119855,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 38901
    },
    {
      "epoch": 15.542149420695166,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 38902
    },
    {
      "epoch": 15.542548941270475,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 38903
    },
    {
      "epoch": 15.542948461845786,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 38904
    },
    {
      "epoch": 15.543347982421094,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 38905
    },
    {
      "epoch": 15.543747502996403,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 38906
    },
    {
      "epoch": 15.544147023571714,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 38907
    },
    {
      "epoch": 15.544546544147023,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 38908
    },
    {
      "epoch": 15.544946064722334,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 38909
    },
    {
      "epoch": 15.545345585297643,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 38910
    },
    {
      "epoch": 15.545745105872953,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0095,
      "step": 38911
    },
    {
      "epoch": 15.546144626448262,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 38912
    },
    {
      "epoch": 15.546544147023571,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 38913
    },
    {
      "epoch": 15.546943667598882,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 38914
    },
    {
      "epoch": 15.54734318817419,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 38915
    },
    {
      "epoch": 15.547742708749501,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 38916
    },
    {
      "epoch": 15.54814222932481,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 38917
    },
    {
      "epoch": 15.548541749900119,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 38918
    },
    {
      "epoch": 15.54894127047543,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 38919
    },
    {
      "epoch": 15.549340791050739,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 38920
    },
    {
      "epoch": 15.54974031162605,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 38921
    },
    {
      "epoch": 15.550139832201358,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 38922
    },
    {
      "epoch": 15.550539352776667,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 38923
    },
    {
      "epoch": 15.550938873351978,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 38924
    },
    {
      "epoch": 15.551338393927287,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 38925
    },
    {
      "epoch": 15.551737914502597,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 38926
    },
    {
      "epoch": 15.552137435077906,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 38927
    },
    {
      "epoch": 15.552536955653217,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 38928
    },
    {
      "epoch": 15.552936476228526,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 38929
    },
    {
      "epoch": 15.553335996803835,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 38930
    },
    {
      "epoch": 15.553735517379145,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 38931
    },
    {
      "epoch": 15.554135037954454,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 38932
    },
    {
      "epoch": 15.554534558529765,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 38933
    },
    {
      "epoch": 15.554934079105074,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 38934
    },
    {
      "epoch": 15.555333599680383,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 38935
    },
    {
      "epoch": 15.555733120255693,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 38936
    },
    {
      "epoch": 15.556132640831002,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 38937
    },
    {
      "epoch": 15.556532161406313,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 38938
    },
    {
      "epoch": 15.556931681981622,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 38939
    },
    {
      "epoch": 15.55733120255693,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 38940
    },
    {
      "epoch": 15.557730723132241,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 38941
    },
    {
      "epoch": 15.55813024370755,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 38942
    },
    {
      "epoch": 15.558529764282861,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 38943
    },
    {
      "epoch": 15.55892928485817,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 38944
    },
    {
      "epoch": 15.55932880543348,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 38945
    },
    {
      "epoch": 15.55972832600879,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 38946
    },
    {
      "epoch": 15.560127846584098,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 38947
    },
    {
      "epoch": 15.560527367159409,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 38948
    },
    {
      "epoch": 15.560926887734718,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 38949
    },
    {
      "epoch": 15.561326408310029,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 38950
    },
    {
      "epoch": 15.561725928885338,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 38951
    },
    {
      "epoch": 15.562125449460646,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 38952
    },
    {
      "epoch": 15.562524970035957,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 38953
    },
    {
      "epoch": 15.562924490611266,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 38954
    },
    {
      "epoch": 15.563324011186577,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 38955
    },
    {
      "epoch": 15.563723531761886,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 38956
    },
    {
      "epoch": 15.564123052337195,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 38957
    },
    {
      "epoch": 15.564522572912505,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 38958
    },
    {
      "epoch": 15.564922093487814,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 38959
    },
    {
      "epoch": 15.565321614063125,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 38960
    },
    {
      "epoch": 15.565721134638434,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 38961
    },
    {
      "epoch": 15.566120655213744,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 38962
    },
    {
      "epoch": 15.566520175789053,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 38963
    },
    {
      "epoch": 15.566919696364362,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 38964
    },
    {
      "epoch": 15.567319216939673,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 38965
    },
    {
      "epoch": 15.567718737514982,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 38966
    },
    {
      "epoch": 15.568118258090292,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 38967
    },
    {
      "epoch": 15.568517778665601,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 38968
    },
    {
      "epoch": 15.56891729924091,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 38969
    },
    {
      "epoch": 15.56931681981622,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 38970
    },
    {
      "epoch": 15.56971634039153,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 38971
    },
    {
      "epoch": 15.57011586096684,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 38972
    },
    {
      "epoch": 15.57051538154215,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 38973
    },
    {
      "epoch": 15.570914902117458,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0433,
      "step": 38974
    },
    {
      "epoch": 15.571314422692769,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 38975
    },
    {
      "epoch": 15.571713943268078,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 38976
    },
    {
      "epoch": 15.572113463843388,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 38977
    },
    {
      "epoch": 15.572512984418697,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 38978
    },
    {
      "epoch": 15.572912504994008,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 38979
    },
    {
      "epoch": 15.573312025569317,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 38980
    },
    {
      "epoch": 15.573711546144626,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 38981
    },
    {
      "epoch": 15.574111066719937,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 38982
    },
    {
      "epoch": 15.574510587295245,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 38983
    },
    {
      "epoch": 15.574910107870556,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 38984
    },
    {
      "epoch": 15.575309628445865,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 38985
    },
    {
      "epoch": 15.575709149021174,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 38986
    },
    {
      "epoch": 15.576108669596485,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 38987
    },
    {
      "epoch": 15.576508190171793,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 38988
    },
    {
      "epoch": 15.576907710747104,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 38989
    },
    {
      "epoch": 15.577307231322413,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 38990
    },
    {
      "epoch": 15.577706751897722,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 38991
    },
    {
      "epoch": 15.578106272473033,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 38992
    },
    {
      "epoch": 15.578505793048341,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 38993
    },
    {
      "epoch": 15.578905313623652,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 38994
    },
    {
      "epoch": 15.579304834198961,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 38995
    },
    {
      "epoch": 15.579704354774272,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 38996
    },
    {
      "epoch": 15.58010387534958,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 38997
    },
    {
      "epoch": 15.58050339592489,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 38998
    },
    {
      "epoch": 15.5809029165002,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 38999
    },
    {
      "epoch": 15.581302437075509,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 39000
    },
    {
      "epoch": 15.58170195765082,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 39001
    },
    {
      "epoch": 15.582101478226129,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 39002
    },
    {
      "epoch": 15.582500998801438,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 39003
    },
    {
      "epoch": 15.582900519376748,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 39004
    },
    {
      "epoch": 15.583300039952057,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 39005
    },
    {
      "epoch": 15.583699560527368,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 39006
    },
    {
      "epoch": 15.584099081102677,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 39007
    },
    {
      "epoch": 15.584498601677986,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 39008
    },
    {
      "epoch": 15.584898122253296,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 39009
    },
    {
      "epoch": 15.585297642828605,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 39010
    },
    {
      "epoch": 15.585697163403916,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 39011
    },
    {
      "epoch": 15.586096683979225,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 39012
    },
    {
      "epoch": 15.586496204554535,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 39013
    },
    {
      "epoch": 15.586895725129844,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 39014
    },
    {
      "epoch": 15.587295245705153,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 39015
    },
    {
      "epoch": 15.587694766280464,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 39016
    },
    {
      "epoch": 15.588094286855773,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 39017
    },
    {
      "epoch": 15.588493807431083,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 39018
    },
    {
      "epoch": 15.588893328006392,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 39019
    },
    {
      "epoch": 15.589292848581701,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 39020
    },
    {
      "epoch": 15.589692369157012,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 39021
    },
    {
      "epoch": 15.59009188973232,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 39022
    },
    {
      "epoch": 15.590491410307632,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 39023
    },
    {
      "epoch": 15.59089093088294,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0104,
      "step": 39024
    },
    {
      "epoch": 15.59129045145825,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 39025
    },
    {
      "epoch": 15.59168997203356,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 39026
    },
    {
      "epoch": 15.592089492608869,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 39027
    },
    {
      "epoch": 15.59248901318418,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 39028
    },
    {
      "epoch": 15.592888533759488,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 39029
    },
    {
      "epoch": 15.5932880543348,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 39030
    },
    {
      "epoch": 15.593687574910108,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 39031
    },
    {
      "epoch": 15.594087095485417,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0443,
      "step": 39032
    },
    {
      "epoch": 15.594486616060728,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 39033
    },
    {
      "epoch": 15.594886136636037,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 39034
    },
    {
      "epoch": 15.595285657211347,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 39035
    },
    {
      "epoch": 15.595685177786656,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 39036
    },
    {
      "epoch": 15.596084698361965,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 39037
    },
    {
      "epoch": 15.596484218937276,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 39038
    },
    {
      "epoch": 15.596883739512585,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 39039
    },
    {
      "epoch": 15.597283260087895,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 39040
    },
    {
      "epoch": 15.597682780663204,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 39041
    },
    {
      "epoch": 15.598082301238513,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 39042
    },
    {
      "epoch": 15.598481821813824,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 39043
    },
    {
      "epoch": 15.598881342389133,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 39044
    },
    {
      "epoch": 15.599280862964443,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 39045
    },
    {
      "epoch": 15.599680383539752,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0107,
      "step": 39046
    },
    {
      "epoch": 15.600079904115063,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 39047
    },
    {
      "epoch": 15.600479424690372,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 39048
    },
    {
      "epoch": 15.60087894526568,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 39049
    },
    {
      "epoch": 15.601278465840991,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 39050
    },
    {
      "epoch": 15.6016779864163,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 39051
    },
    {
      "epoch": 15.602077506991611,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 39052
    },
    {
      "epoch": 15.60247702756692,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 39053
    },
    {
      "epoch": 15.602876548142229,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 39054
    },
    {
      "epoch": 15.60327606871754,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 39055
    },
    {
      "epoch": 15.603675589292848,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 39056
    },
    {
      "epoch": 15.604075109868159,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 39057
    },
    {
      "epoch": 15.604474630443468,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 39058
    },
    {
      "epoch": 15.604874151018777,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 39059
    },
    {
      "epoch": 15.605273671594087,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 39060
    },
    {
      "epoch": 15.605673192169396,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 39061
    },
    {
      "epoch": 15.606072712744707,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 39062
    },
    {
      "epoch": 15.606472233320016,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 39063
    },
    {
      "epoch": 15.606871753895327,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 39064
    },
    {
      "epoch": 15.607271274470635,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 39065
    },
    {
      "epoch": 15.607670795045944,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.008,
      "step": 39066
    },
    {
      "epoch": 15.608070315621255,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 39067
    },
    {
      "epoch": 15.608469836196564,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 39068
    },
    {
      "epoch": 15.608869356771873,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 39069
    },
    {
      "epoch": 15.609268877347183,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 39070
    },
    {
      "epoch": 15.609668397922492,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 39071
    },
    {
      "epoch": 15.610067918497803,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 39072
    },
    {
      "epoch": 15.610467439073112,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 39073
    },
    {
      "epoch": 15.610866959648423,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 39074
    },
    {
      "epoch": 15.611266480223732,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 39075
    },
    {
      "epoch": 15.61166600079904,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 39076
    },
    {
      "epoch": 15.612065521374351,
      "grad_norm": 0.036865234375,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 39077
    },
    {
      "epoch": 15.61246504194966,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 39078
    },
    {
      "epoch": 15.61286456252497,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 39079
    },
    {
      "epoch": 15.61326408310028,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 39080
    },
    {
      "epoch": 15.613663603675588,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 39081
    },
    {
      "epoch": 15.6140631242509,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 39082
    },
    {
      "epoch": 15.614462644826208,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 39083
    },
    {
      "epoch": 15.614862165401519,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 39084
    },
    {
      "epoch": 15.615261685976828,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 39085
    },
    {
      "epoch": 15.615661206552137,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 39086
    },
    {
      "epoch": 15.616060727127447,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 39087
    },
    {
      "epoch": 15.616460247702756,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 39088
    },
    {
      "epoch": 15.616859768278067,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 39089
    },
    {
      "epoch": 15.617259288853376,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 39090
    },
    {
      "epoch": 15.617658809428686,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 39091
    },
    {
      "epoch": 15.618058330003995,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 39092
    },
    {
      "epoch": 15.618457850579304,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0128,
      "step": 39093
    },
    {
      "epoch": 15.618857371154615,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 39094
    },
    {
      "epoch": 15.619256891729924,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 39095
    },
    {
      "epoch": 15.619656412305234,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 39096
    },
    {
      "epoch": 15.620055932880543,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 39097
    },
    {
      "epoch": 15.620455453455852,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 39098
    },
    {
      "epoch": 15.620854974031163,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 39099
    },
    {
      "epoch": 15.621254494606472,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 39100
    },
    {
      "epoch": 15.621654015181782,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 39101
    },
    {
      "epoch": 15.622053535757091,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 39102
    },
    {
      "epoch": 15.6224530563324,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 39103
    },
    {
      "epoch": 15.622852576907711,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 39104
    },
    {
      "epoch": 15.62325209748302,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 39105
    },
    {
      "epoch": 15.62365161805833,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 39106
    },
    {
      "epoch": 15.62405113863364,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 39107
    },
    {
      "epoch": 15.62445065920895,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 39108
    },
    {
      "epoch": 15.624850179784259,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 39109
    },
    {
      "epoch": 15.625249700359568,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 39110
    },
    {
      "epoch": 15.625649220934879,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0101,
      "step": 39111
    },
    {
      "epoch": 15.626048741510187,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 39112
    },
    {
      "epoch": 15.626448262085498,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 39113
    },
    {
      "epoch": 15.626847782660807,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 39114
    },
    {
      "epoch": 15.627247303236116,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 39115
    },
    {
      "epoch": 15.627646823811427,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 39116
    },
    {
      "epoch": 15.628046344386735,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 39117
    },
    {
      "epoch": 15.628445864962046,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 39118
    },
    {
      "epoch": 15.628845385537355,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 39119
    },
    {
      "epoch": 15.629244906112664,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 39120
    },
    {
      "epoch": 15.629644426687975,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 39121
    },
    {
      "epoch": 15.630043947263283,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 39122
    },
    {
      "epoch": 15.630443467838594,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 39123
    },
    {
      "epoch": 15.630842988413903,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 39124
    },
    {
      "epoch": 15.631242508989214,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 39125
    },
    {
      "epoch": 15.631642029564523,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 39126
    },
    {
      "epoch": 15.632041550139832,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 39127
    },
    {
      "epoch": 15.632441070715142,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 39128
    },
    {
      "epoch": 15.632840591290451,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 39129
    },
    {
      "epoch": 15.633240111865762,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 39130
    },
    {
      "epoch": 15.63363963244107,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0114,
      "step": 39131
    },
    {
      "epoch": 15.63403915301638,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 39132
    },
    {
      "epoch": 15.63443867359169,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 39133
    },
    {
      "epoch": 15.634838194167,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 39134
    },
    {
      "epoch": 15.63523771474231,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 39135
    },
    {
      "epoch": 15.635637235317619,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 39136
    },
    {
      "epoch": 15.636036755892928,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 39137
    },
    {
      "epoch": 15.636436276468238,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 39138
    },
    {
      "epoch": 15.636835797043547,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 39139
    },
    {
      "epoch": 15.637235317618858,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 39140
    },
    {
      "epoch": 15.637634838194167,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 39141
    },
    {
      "epoch": 15.638034358769477,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 39142
    },
    {
      "epoch": 15.638433879344786,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 39143
    },
    {
      "epoch": 15.638833399920095,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 39144
    },
    {
      "epoch": 15.639232920495406,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 39145
    },
    {
      "epoch": 15.639632441070715,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 39146
    },
    {
      "epoch": 15.640031961646025,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 39147
    },
    {
      "epoch": 15.640431482221334,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 39148
    },
    {
      "epoch": 15.640831002796643,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 39149
    },
    {
      "epoch": 15.641230523371954,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 39150
    },
    {
      "epoch": 15.641630043947263,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 39151
    },
    {
      "epoch": 15.642029564522574,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 39152
    },
    {
      "epoch": 15.642429085097882,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 39153
    },
    {
      "epoch": 15.642828605673191,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 39154
    },
    {
      "epoch": 15.643228126248502,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 39155
    },
    {
      "epoch": 15.643627646823811,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 39156
    },
    {
      "epoch": 15.644027167399122,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 39157
    },
    {
      "epoch": 15.64442668797443,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 39158
    },
    {
      "epoch": 15.644826208549741,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 39159
    },
    {
      "epoch": 15.64522572912505,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 39160
    },
    {
      "epoch": 15.645625249700359,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 39161
    },
    {
      "epoch": 15.64602477027567,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 39162
    },
    {
      "epoch": 15.646424290850979,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 39163
    },
    {
      "epoch": 15.64682381142629,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 39164
    },
    {
      "epoch": 15.647223332001598,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 39165
    },
    {
      "epoch": 15.647622852576907,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 39166
    },
    {
      "epoch": 15.648022373152218,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 39167
    },
    {
      "epoch": 15.648421893727527,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 39168
    },
    {
      "epoch": 15.648821414302837,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 39169
    },
    {
      "epoch": 15.649220934878146,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 39170
    },
    {
      "epoch": 15.649620455453455,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 39171
    },
    {
      "epoch": 15.650019976028766,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 39172
    },
    {
      "epoch": 15.650419496604075,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 39173
    },
    {
      "epoch": 15.650819017179385,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 39174
    },
    {
      "epoch": 15.651218537754694,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 39175
    },
    {
      "epoch": 15.651618058330005,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 39176
    },
    {
      "epoch": 15.652017578905314,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 39177
    },
    {
      "epoch": 15.652417099480623,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 39178
    },
    {
      "epoch": 15.652816620055933,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 39179
    },
    {
      "epoch": 15.653216140631242,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 39180
    },
    {
      "epoch": 15.653615661206553,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 39181
    },
    {
      "epoch": 15.654015181781862,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0474,
      "step": 39182
    },
    {
      "epoch": 15.65441470235717,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 39183
    },
    {
      "epoch": 15.654814222932481,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 39184
    },
    {
      "epoch": 15.65521374350779,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 39185
    },
    {
      "epoch": 15.655613264083101,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 39186
    },
    {
      "epoch": 15.65601278465841,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 39187
    },
    {
      "epoch": 15.656412305233719,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 39188
    },
    {
      "epoch": 15.65681182580903,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 39189
    },
    {
      "epoch": 15.657211346384338,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 39190
    },
    {
      "epoch": 15.657610866959649,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 39191
    },
    {
      "epoch": 15.658010387534958,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 39192
    },
    {
      "epoch": 15.658409908110269,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 39193
    },
    {
      "epoch": 15.658809428685577,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 39194
    },
    {
      "epoch": 15.659208949260886,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 39195
    },
    {
      "epoch": 15.659608469836197,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 39196
    },
    {
      "epoch": 15.660007990411506,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 39197
    },
    {
      "epoch": 15.660407510986817,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 39198
    },
    {
      "epoch": 15.660807031562126,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 39199
    },
    {
      "epoch": 15.661206552137434,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 39200
    },
    {
      "epoch": 15.661606072712745,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 39201
    },
    {
      "epoch": 15.662005593288054,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 39202
    },
    {
      "epoch": 15.662405113863365,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 39203
    },
    {
      "epoch": 15.662804634438674,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 39204
    },
    {
      "epoch": 15.663204155013982,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 39205
    },
    {
      "epoch": 15.663603675589293,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 39206
    },
    {
      "epoch": 15.664003196164602,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 39207
    },
    {
      "epoch": 15.664402716739913,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 39208
    },
    {
      "epoch": 15.664802237315222,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 39209
    },
    {
      "epoch": 15.665201757890532,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 39210
    },
    {
      "epoch": 15.665601278465841,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 39211
    },
    {
      "epoch": 15.66600079904115,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 39212
    },
    {
      "epoch": 15.66640031961646,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 39213
    },
    {
      "epoch": 15.66679984019177,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 39214
    },
    {
      "epoch": 15.66719936076708,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 39215
    },
    {
      "epoch": 15.66759888134239,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 39216
    },
    {
      "epoch": 15.667998401917698,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 39217
    },
    {
      "epoch": 15.668397922493009,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 39218
    },
    {
      "epoch": 15.668797443068318,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 39219
    },
    {
      "epoch": 15.669196963643628,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 39220
    },
    {
      "epoch": 15.669596484218937,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 39221
    },
    {
      "epoch": 15.669996004794246,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0453,
      "step": 39222
    },
    {
      "epoch": 15.670395525369557,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 39223
    },
    {
      "epoch": 15.670795045944866,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 39224
    },
    {
      "epoch": 15.671194566520176,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 39225
    },
    {
      "epoch": 15.671594087095485,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 39226
    },
    {
      "epoch": 15.671993607670796,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 39227
    },
    {
      "epoch": 15.672393128246105,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 39228
    },
    {
      "epoch": 15.672792648821414,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 39229
    },
    {
      "epoch": 15.673192169396724,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 39230
    },
    {
      "epoch": 15.673591689972033,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 39231
    },
    {
      "epoch": 15.673991210547344,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 39232
    },
    {
      "epoch": 15.674390731122653,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 39233
    },
    {
      "epoch": 15.674790251697962,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 39234
    },
    {
      "epoch": 15.675189772273272,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 39235
    },
    {
      "epoch": 15.675589292848581,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 39236
    },
    {
      "epoch": 15.675988813423892,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 39237
    },
    {
      "epoch": 15.676388333999201,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 39238
    },
    {
      "epoch": 15.67678785457451,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 39239
    },
    {
      "epoch": 15.67718737514982,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 39240
    },
    {
      "epoch": 15.67758689572513,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 39241
    },
    {
      "epoch": 15.67798641630044,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0118,
      "step": 39242
    },
    {
      "epoch": 15.678385936875749,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 39243
    },
    {
      "epoch": 15.67878545745106,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 39244
    },
    {
      "epoch": 15.679184978026369,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 39245
    },
    {
      "epoch": 15.679584498601677,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 39246
    },
    {
      "epoch": 15.679984019176988,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 39247
    },
    {
      "epoch": 15.680383539752297,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 39248
    },
    {
      "epoch": 15.680783060327606,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 39249
    },
    {
      "epoch": 15.681182580902917,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 39250
    },
    {
      "epoch": 15.681582101478226,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 39251
    },
    {
      "epoch": 15.681981622053536,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 39252
    },
    {
      "epoch": 15.682381142628845,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 39253
    },
    {
      "epoch": 15.682780663204156,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 39254
    },
    {
      "epoch": 15.683180183779465,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 39255
    },
    {
      "epoch": 15.683579704354774,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 39256
    },
    {
      "epoch": 15.683979224930084,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 39257
    },
    {
      "epoch": 15.684378745505393,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 39258
    },
    {
      "epoch": 15.684778266080704,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 39259
    },
    {
      "epoch": 15.685177786656013,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 39260
    },
    {
      "epoch": 15.685577307231322,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 39261
    },
    {
      "epoch": 15.685976827806632,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 39262
    },
    {
      "epoch": 15.686376348381941,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 39263
    },
    {
      "epoch": 15.686775868957252,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 39264
    },
    {
      "epoch": 15.68717538953256,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 39265
    },
    {
      "epoch": 15.68757491010787,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 39266
    },
    {
      "epoch": 15.68797443068318,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 39267
    },
    {
      "epoch": 15.68837395125849,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 39268
    },
    {
      "epoch": 15.6887734718338,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 39269
    },
    {
      "epoch": 15.689172992409109,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 39270
    },
    {
      "epoch": 15.68957251298442,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 39271
    },
    {
      "epoch": 15.689972033559728,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 39272
    },
    {
      "epoch": 15.690371554135037,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 39273
    },
    {
      "epoch": 15.690771074710348,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 39274
    },
    {
      "epoch": 15.691170595285657,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 39275
    },
    {
      "epoch": 15.691570115860968,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 39276
    },
    {
      "epoch": 15.691969636436276,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 39277
    },
    {
      "epoch": 15.692369157011585,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 39278
    },
    {
      "epoch": 15.692768677586896,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 39279
    },
    {
      "epoch": 15.693168198162205,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 39280
    },
    {
      "epoch": 15.693567718737516,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 39281
    },
    {
      "epoch": 15.693967239312824,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 39282
    },
    {
      "epoch": 15.694366759888133,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 39283
    },
    {
      "epoch": 15.694766280463444,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 39284
    },
    {
      "epoch": 15.695165801038753,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 39285
    },
    {
      "epoch": 15.695565321614064,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 39286
    },
    {
      "epoch": 15.695964842189372,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 39287
    },
    {
      "epoch": 15.696364362764683,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 39288
    },
    {
      "epoch": 15.696763883339992,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 39289
    },
    {
      "epoch": 15.697163403915301,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 39290
    },
    {
      "epoch": 15.697562924490612,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 39291
    },
    {
      "epoch": 15.69796244506592,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 39292
    },
    {
      "epoch": 15.698361965641231,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 39293
    },
    {
      "epoch": 15.69876148621654,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 39294
    },
    {
      "epoch": 15.699161006791849,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 39295
    },
    {
      "epoch": 15.69956052736716,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 39296
    },
    {
      "epoch": 15.699960047942469,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 39297
    },
    {
      "epoch": 15.70035956851778,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 39298
    },
    {
      "epoch": 15.700759089093088,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 39299
    },
    {
      "epoch": 15.701158609668397,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 39300
    },
    {
      "epoch": 15.701558130243708,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 39301
    },
    {
      "epoch": 15.701957650819017,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 39302
    },
    {
      "epoch": 15.702357171394327,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 39303
    },
    {
      "epoch": 15.702756691969636,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 39304
    },
    {
      "epoch": 15.703156212544947,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 39305
    },
    {
      "epoch": 15.703555733120256,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 39306
    },
    {
      "epoch": 15.703955253695565,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 39307
    },
    {
      "epoch": 15.704354774270875,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 39308
    },
    {
      "epoch": 15.704754294846184,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 39309
    },
    {
      "epoch": 15.705153815421495,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 39310
    },
    {
      "epoch": 15.705553335996804,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 39311
    },
    {
      "epoch": 15.705952856572113,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 39312
    },
    {
      "epoch": 15.706352377147423,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0005,
      "loss": 1.0122,
      "step": 39313
    },
    {
      "epoch": 15.706751897722732,
      "grad_norm": 0.265625,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 39314
    },
    {
      "epoch": 15.707151418298043,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 39315
    },
    {
      "epoch": 15.707550938873352,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 39316
    },
    {
      "epoch": 15.70795045944866,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 39317
    },
    {
      "epoch": 15.708349980023971,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 39318
    },
    {
      "epoch": 15.70874950059928,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 39319
    },
    {
      "epoch": 15.709149021174591,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 39320
    },
    {
      "epoch": 15.7095485417499,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 39321
    },
    {
      "epoch": 15.70994806232521,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 39322
    },
    {
      "epoch": 15.71034758290052,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 39323
    },
    {
      "epoch": 15.710747103475828,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 39324
    },
    {
      "epoch": 15.711146624051139,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 39325
    },
    {
      "epoch": 15.711546144626448,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 39326
    },
    {
      "epoch": 15.711945665201759,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 39327
    },
    {
      "epoch": 15.712345185777068,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 39328
    },
    {
      "epoch": 15.712744706352376,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 39329
    },
    {
      "epoch": 15.713144226927687,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 39330
    },
    {
      "epoch": 15.713543747502996,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 39331
    },
    {
      "epoch": 15.713943268078307,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 39332
    },
    {
      "epoch": 15.714342788653616,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 39333
    },
    {
      "epoch": 15.714742309228924,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 39334
    },
    {
      "epoch": 15.715141829804235,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 39335
    },
    {
      "epoch": 15.715541350379544,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0126,
      "step": 39336
    },
    {
      "epoch": 15.715940870954855,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 39337
    },
    {
      "epoch": 15.716340391530164,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 39338
    },
    {
      "epoch": 15.716739912105474,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 39339
    },
    {
      "epoch": 15.717139432680783,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 39340
    },
    {
      "epoch": 15.717538953256092,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 39341
    },
    {
      "epoch": 15.717938473831403,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 39342
    },
    {
      "epoch": 15.718337994406712,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 39343
    },
    {
      "epoch": 15.718737514982022,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 39344
    },
    {
      "epoch": 15.719137035557331,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 39345
    },
    {
      "epoch": 15.71953655613264,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 39346
    },
    {
      "epoch": 15.71993607670795,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 39347
    },
    {
      "epoch": 15.72033559728326,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 39348
    },
    {
      "epoch": 15.72073511785857,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 39349
    },
    {
      "epoch": 15.72113463843388,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 39350
    },
    {
      "epoch": 15.721534159009188,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 39351
    },
    {
      "epoch": 15.721933679584499,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 39352
    },
    {
      "epoch": 15.722333200159808,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 39353
    },
    {
      "epoch": 15.722732720735118,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 39354
    },
    {
      "epoch": 15.723132241310427,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 39355
    },
    {
      "epoch": 15.723531761885738,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 39356
    },
    {
      "epoch": 15.723931282461047,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 39357
    },
    {
      "epoch": 15.724330803036356,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 39358
    },
    {
      "epoch": 15.724730323611666,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 39359
    },
    {
      "epoch": 15.725129844186975,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0114,
      "step": 39360
    },
    {
      "epoch": 15.725529364762286,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 39361
    },
    {
      "epoch": 15.725928885337595,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 39362
    },
    {
      "epoch": 15.726328405912904,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 39363
    },
    {
      "epoch": 15.726727926488214,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 39364
    },
    {
      "epoch": 15.727127447063523,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 39365
    },
    {
      "epoch": 15.727526967638834,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 39366
    },
    {
      "epoch": 15.727926488214143,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 39367
    },
    {
      "epoch": 15.728326008789452,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 39368
    },
    {
      "epoch": 15.728725529364763,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 39369
    },
    {
      "epoch": 15.729125049940071,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 39370
    },
    {
      "epoch": 15.729524570515382,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 39371
    },
    {
      "epoch": 15.729924091090691,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 39372
    },
    {
      "epoch": 15.730323611666002,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 39373
    },
    {
      "epoch": 15.73072313224131,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 39374
    },
    {
      "epoch": 15.73112265281662,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 39375
    },
    {
      "epoch": 15.73152217339193,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 39376
    },
    {
      "epoch": 15.731921693967239,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 39377
    },
    {
      "epoch": 15.73232121454255,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 39378
    },
    {
      "epoch": 15.732720735117859,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 39379
    },
    {
      "epoch": 15.733120255693168,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 39380
    },
    {
      "epoch": 15.733519776268478,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 39381
    },
    {
      "epoch": 15.733919296843787,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 39382
    },
    {
      "epoch": 15.734318817419098,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 39383
    },
    {
      "epoch": 15.734718337994407,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 39384
    },
    {
      "epoch": 15.735117858569716,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 39385
    },
    {
      "epoch": 15.735517379145026,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 39386
    },
    {
      "epoch": 15.735916899720335,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 39387
    },
    {
      "epoch": 15.736316420295646,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 39388
    },
    {
      "epoch": 15.736715940870955,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 39389
    },
    {
      "epoch": 15.737115461446265,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 39390
    },
    {
      "epoch": 15.737514982021574,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 39391
    },
    {
      "epoch": 15.737914502596883,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 39392
    },
    {
      "epoch": 15.738314023172194,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 39393
    },
    {
      "epoch": 15.738713543747503,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 39394
    },
    {
      "epoch": 15.739113064322813,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 39395
    },
    {
      "epoch": 15.739512584898122,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 39396
    },
    {
      "epoch": 15.739912105473431,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 39397
    },
    {
      "epoch": 15.740311626048742,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 39398
    },
    {
      "epoch": 15.74071114662405,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 39399
    },
    {
      "epoch": 15.741110667199361,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 39400
    },
    {
      "epoch": 15.74151018777467,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 39401
    },
    {
      "epoch": 15.74190970834998,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 39402
    },
    {
      "epoch": 15.74230922892529,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 39403
    },
    {
      "epoch": 15.742708749500599,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 39404
    },
    {
      "epoch": 15.74310827007591,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 39405
    },
    {
      "epoch": 15.743507790651218,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 39406
    },
    {
      "epoch": 15.743907311226529,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 39407
    },
    {
      "epoch": 15.744306831801838,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 39408
    },
    {
      "epoch": 15.744706352377147,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 39409
    },
    {
      "epoch": 15.745105872952458,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 39410
    },
    {
      "epoch": 15.745505393527766,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 39411
    },
    {
      "epoch": 15.745904914103077,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 39412
    },
    {
      "epoch": 15.746304434678386,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 39413
    },
    {
      "epoch": 15.746703955253695,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 39414
    },
    {
      "epoch": 15.747103475829006,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0131,
      "step": 39415
    },
    {
      "epoch": 15.747502996404315,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 39416
    },
    {
      "epoch": 15.747902516979625,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 39417
    },
    {
      "epoch": 15.748302037554934,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 39418
    },
    {
      "epoch": 15.748701558130243,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 39419
    },
    {
      "epoch": 15.749101078705554,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 39420
    },
    {
      "epoch": 15.749500599280863,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 39421
    },
    {
      "epoch": 15.749900119856173,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 39422
    },
    {
      "epoch": 15.750299640431482,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 39423
    },
    {
      "epoch": 15.750699161006793,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 39424
    },
    {
      "epoch": 15.751098681582102,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 39425
    },
    {
      "epoch": 15.75149820215741,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 39426
    },
    {
      "epoch": 15.751897722732721,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 39427
    },
    {
      "epoch": 15.75229724330803,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 39428
    },
    {
      "epoch": 15.752696763883339,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 39429
    },
    {
      "epoch": 15.75309628445865,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 39430
    },
    {
      "epoch": 15.753495805033959,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 39431
    },
    {
      "epoch": 15.75389532560927,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 39432
    },
    {
      "epoch": 15.754294846184578,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 39433
    },
    {
      "epoch": 15.754694366759889,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 39434
    },
    {
      "epoch": 15.755093887335198,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 39435
    },
    {
      "epoch": 15.755493407910507,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 39436
    },
    {
      "epoch": 15.755892928485817,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 39437
    },
    {
      "epoch": 15.756292449061126,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 39438
    },
    {
      "epoch": 15.756691969636437,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 39439
    },
    {
      "epoch": 15.757091490211746,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 39440
    },
    {
      "epoch": 15.757491010787055,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 39441
    },
    {
      "epoch": 15.757890531362365,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 39442
    },
    {
      "epoch": 15.758290051937674,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 39443
    },
    {
      "epoch": 15.758689572512985,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 39444
    },
    {
      "epoch": 15.759089093088294,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 39445
    },
    {
      "epoch": 15.759488613663603,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 39446
    },
    {
      "epoch": 15.759888134238913,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 39447
    },
    {
      "epoch": 15.760287654814222,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 39448
    },
    {
      "epoch": 15.760687175389533,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 39449
    },
    {
      "epoch": 15.761086695964842,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 39450
    },
    {
      "epoch": 15.761486216540153,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 39451
    },
    {
      "epoch": 15.761885737115461,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 39452
    },
    {
      "epoch": 15.76228525769077,
      "grad_norm": 0.036865234375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 39453
    },
    {
      "epoch": 15.762684778266081,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 39454
    },
    {
      "epoch": 15.76308429884139,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 39455
    },
    {
      "epoch": 15.7634838194167,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 39456
    },
    {
      "epoch": 15.76388333999201,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 39457
    },
    {
      "epoch": 15.764282860567318,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 39458
    },
    {
      "epoch": 15.764682381142629,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 39459
    },
    {
      "epoch": 15.765081901717938,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 39460
    },
    {
      "epoch": 15.765481422293249,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 39461
    },
    {
      "epoch": 15.765880942868558,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 39462
    },
    {
      "epoch": 15.766280463443866,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 39463
    },
    {
      "epoch": 15.766679984019177,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 39464
    },
    {
      "epoch": 15.767079504594486,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 39465
    },
    {
      "epoch": 15.767479025169797,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 39466
    },
    {
      "epoch": 15.767878545745106,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0127,
      "step": 39467
    },
    {
      "epoch": 15.768278066320416,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 39468
    },
    {
      "epoch": 15.768677586895725,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 39469
    },
    {
      "epoch": 15.769077107471034,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 39470
    },
    {
      "epoch": 15.769476628046345,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 39471
    },
    {
      "epoch": 15.769876148621654,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 39472
    },
    {
      "epoch": 15.770275669196964,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 39473
    },
    {
      "epoch": 15.770675189772273,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 39474
    },
    {
      "epoch": 15.771074710347582,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 39475
    },
    {
      "epoch": 15.771474230922893,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 39476
    },
    {
      "epoch": 15.771873751498202,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 39477
    },
    {
      "epoch": 15.772273272073512,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 39478
    },
    {
      "epoch": 15.772672792648821,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 39479
    },
    {
      "epoch": 15.77307231322413,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 39480
    },
    {
      "epoch": 15.77347183379944,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 39481
    },
    {
      "epoch": 15.77387135437475,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 39482
    },
    {
      "epoch": 15.77427087495006,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 39483
    },
    {
      "epoch": 15.77467039552537,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 39484
    },
    {
      "epoch": 15.77506991610068,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 39485
    },
    {
      "epoch": 15.775469436675989,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 39486
    },
    {
      "epoch": 15.775868957251298,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 39487
    },
    {
      "epoch": 15.776268477826608,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 39488
    },
    {
      "epoch": 15.776667998401917,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 39489
    },
    {
      "epoch": 15.777067518977228,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 39490
    },
    {
      "epoch": 15.777467039552537,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 39491
    },
    {
      "epoch": 15.777866560127846,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 39492
    },
    {
      "epoch": 15.778266080703157,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 39493
    },
    {
      "epoch": 15.778665601278465,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 39494
    },
    {
      "epoch": 15.779065121853776,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 39495
    },
    {
      "epoch": 15.779464642429085,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 39496
    },
    {
      "epoch": 15.779864163004394,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 39497
    },
    {
      "epoch": 15.780263683579705,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 39498
    },
    {
      "epoch": 15.780663204155013,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 39499
    },
    {
      "epoch": 15.781062724730324,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 39500
    },
    {
      "epoch": 15.781462245305633,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 39501
    },
    {
      "epoch": 15.781861765880944,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 39502
    },
    {
      "epoch": 15.782261286456253,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 39503
    },
    {
      "epoch": 15.782660807031561,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 39504
    },
    {
      "epoch": 15.783060327606872,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 39505
    },
    {
      "epoch": 15.783459848182181,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 39506
    },
    {
      "epoch": 15.783859368757492,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 39507
    },
    {
      "epoch": 15.7842588893328,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 39508
    },
    {
      "epoch": 15.78465840990811,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 39509
    },
    {
      "epoch": 15.78505793048342,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 39510
    },
    {
      "epoch": 15.785457451058729,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 39511
    },
    {
      "epoch": 15.78585697163404,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 39512
    },
    {
      "epoch": 15.786256492209349,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 39513
    },
    {
      "epoch": 15.786656012784658,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 39514
    },
    {
      "epoch": 15.787055533359968,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 39515
    },
    {
      "epoch": 15.787455053935277,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 39516
    },
    {
      "epoch": 15.787854574510588,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 39517
    },
    {
      "epoch": 15.788254095085897,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 39518
    },
    {
      "epoch": 15.788653615661207,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 39519
    },
    {
      "epoch": 15.789053136236516,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 39520
    },
    {
      "epoch": 15.789452656811825,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 39521
    },
    {
      "epoch": 15.789852177387136,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 39522
    },
    {
      "epoch": 15.790251697962445,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 39523
    },
    {
      "epoch": 15.790651218537755,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 39524
    },
    {
      "epoch": 15.791050739113064,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 39525
    },
    {
      "epoch": 15.791450259688373,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 39526
    },
    {
      "epoch": 15.791849780263684,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 39527
    },
    {
      "epoch": 15.792249300838993,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 39528
    },
    {
      "epoch": 15.792648821414303,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 39529
    },
    {
      "epoch": 15.793048341989612,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 39530
    },
    {
      "epoch": 15.793447862564921,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 39531
    },
    {
      "epoch": 15.793847383140232,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 39532
    },
    {
      "epoch": 15.79424690371554,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 39533
    },
    {
      "epoch": 15.794646424290852,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 39534
    },
    {
      "epoch": 15.79504594486616,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 39535
    },
    {
      "epoch": 15.795445465441471,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 39536
    },
    {
      "epoch": 15.79584498601678,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 39537
    },
    {
      "epoch": 15.796244506592089,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 39538
    },
    {
      "epoch": 15.7966440271674,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 39539
    },
    {
      "epoch": 15.797043547742708,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 39540
    },
    {
      "epoch": 15.79744306831802,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 39541
    },
    {
      "epoch": 15.797842588893328,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 39542
    },
    {
      "epoch": 15.798242109468637,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 39543
    },
    {
      "epoch": 15.798641630043948,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 39544
    },
    {
      "epoch": 15.799041150619257,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 39545
    },
    {
      "epoch": 15.799440671194567,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 39546
    },
    {
      "epoch": 15.799840191769876,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 39547
    },
    {
      "epoch": 15.800239712345185,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 39548
    },
    {
      "epoch": 15.800639232920496,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 39549
    },
    {
      "epoch": 15.801038753495805,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 39550
    },
    {
      "epoch": 15.801438274071115,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 39551
    },
    {
      "epoch": 15.801837794646424,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 39552
    },
    {
      "epoch": 15.802237315221735,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 39553
    },
    {
      "epoch": 15.802636835797044,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 39554
    },
    {
      "epoch": 15.803036356372353,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 39555
    },
    {
      "epoch": 15.803435876947663,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 39556
    },
    {
      "epoch": 15.803835397522972,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 39557
    },
    {
      "epoch": 15.804234918098283,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 39558
    },
    {
      "epoch": 15.804634438673592,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 39559
    },
    {
      "epoch": 15.8050339592489,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 39560
    },
    {
      "epoch": 15.805433479824211,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 39561
    },
    {
      "epoch": 15.80583300039952,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 39562
    },
    {
      "epoch": 15.806232520974831,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 39563
    },
    {
      "epoch": 15.80663204155014,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 39564
    },
    {
      "epoch": 15.807031562125449,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 39565
    },
    {
      "epoch": 15.80743108270076,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 39566
    },
    {
      "epoch": 15.807830603276068,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 39567
    },
    {
      "epoch": 15.808230123851379,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 39568
    },
    {
      "epoch": 15.808629644426688,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 39569
    },
    {
      "epoch": 15.809029165001999,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 39570
    },
    {
      "epoch": 15.809428685577307,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 39571
    },
    {
      "epoch": 15.809828206152616,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 39572
    },
    {
      "epoch": 15.810227726727927,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 39573
    },
    {
      "epoch": 15.810627247303236,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0125,
      "step": 39574
    },
    {
      "epoch": 15.811026767878547,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 39575
    },
    {
      "epoch": 15.811426288453855,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 39576
    },
    {
      "epoch": 15.811825809029164,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 39577
    },
    {
      "epoch": 15.812225329604475,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 39578
    },
    {
      "epoch": 15.812624850179784,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 39579
    },
    {
      "epoch": 15.813024370755095,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 39580
    },
    {
      "epoch": 15.813423891330403,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 39581
    },
    {
      "epoch": 15.813823411905712,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 39582
    },
    {
      "epoch": 15.814222932481023,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 39583
    },
    {
      "epoch": 15.814622453056332,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 39584
    },
    {
      "epoch": 15.815021973631643,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 39585
    },
    {
      "epoch": 15.815421494206952,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 39586
    },
    {
      "epoch": 15.815821014782262,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 39587
    },
    {
      "epoch": 15.816220535357571,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 39588
    },
    {
      "epoch": 15.81662005593288,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 39589
    },
    {
      "epoch": 15.81701957650819,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 39590
    },
    {
      "epoch": 15.8174190970835,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 39591
    },
    {
      "epoch": 15.81781861765881,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 39592
    },
    {
      "epoch": 15.81821813823412,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 39593
    },
    {
      "epoch": 15.818617658809428,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 39594
    },
    {
      "epoch": 15.819017179384739,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 39595
    },
    {
      "epoch": 15.819416699960048,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 39596
    },
    {
      "epoch": 15.819816220535358,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 39597
    },
    {
      "epoch": 15.820215741110667,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 39598
    },
    {
      "epoch": 15.820615261685976,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 39599
    },
    {
      "epoch": 15.821014782261287,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 39600
    },
    {
      "epoch": 15.821414302836596,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 39601
    },
    {
      "epoch": 15.821813823411906,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 39602
    },
    {
      "epoch": 15.822213343987215,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 39603
    },
    {
      "epoch": 15.822612864562524,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0005,
      "loss": 1.0438,
      "step": 39604
    },
    {
      "epoch": 15.823012385137835,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 39605
    },
    {
      "epoch": 15.823411905713144,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 39606
    },
    {
      "epoch": 15.823811426288454,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 39607
    },
    {
      "epoch": 15.824210946863763,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 39608
    },
    {
      "epoch": 15.824610467439072,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 39609
    },
    {
      "epoch": 15.825009988014383,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 39610
    },
    {
      "epoch": 15.825409508589692,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 39611
    },
    {
      "epoch": 15.825809029165002,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 39612
    },
    {
      "epoch": 15.826208549740311,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0135,
      "step": 39613
    },
    {
      "epoch": 15.826608070315622,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 39614
    },
    {
      "epoch": 15.827007590890931,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 39615
    },
    {
      "epoch": 15.82740711146624,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 39616
    },
    {
      "epoch": 15.82780663204155,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 39617
    },
    {
      "epoch": 15.82820615261686,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 39618
    },
    {
      "epoch": 15.82860567319217,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 39619
    },
    {
      "epoch": 15.829005193767479,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 39620
    },
    {
      "epoch": 15.829404714342788,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 39621
    },
    {
      "epoch": 15.829804234918099,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 39622
    },
    {
      "epoch": 15.830203755493407,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 39623
    },
    {
      "epoch": 15.830603276068718,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 39624
    },
    {
      "epoch": 15.831002796644027,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 39625
    },
    {
      "epoch": 15.831402317219336,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 39626
    },
    {
      "epoch": 15.831801837794647,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 39627
    },
    {
      "epoch": 15.832201358369955,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 39628
    },
    {
      "epoch": 15.832600878945266,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 39629
    },
    {
      "epoch": 15.833000399520575,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 39630
    },
    {
      "epoch": 15.833399920095886,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 39631
    },
    {
      "epoch": 15.833799440671195,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 39632
    },
    {
      "epoch": 15.834198961246504,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 39633
    },
    {
      "epoch": 15.834598481821814,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 39634
    },
    {
      "epoch": 15.834998002397123,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 39635
    },
    {
      "epoch": 15.835397522972434,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0085,
      "step": 39636
    },
    {
      "epoch": 15.835797043547743,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 39637
    },
    {
      "epoch": 15.836196564123052,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 39638
    },
    {
      "epoch": 15.836596084698362,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 39639
    },
    {
      "epoch": 15.836995605273671,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 39640
    },
    {
      "epoch": 15.837395125848982,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 39641
    },
    {
      "epoch": 15.83779464642429,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 39642
    },
    {
      "epoch": 15.8381941669996,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 39643
    },
    {
      "epoch": 15.83859368757491,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 39644
    },
    {
      "epoch": 15.83899320815022,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 39645
    },
    {
      "epoch": 15.83939272872553,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 39646
    },
    {
      "epoch": 15.839792249300839,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 39647
    },
    {
      "epoch": 15.84019176987615,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 39648
    },
    {
      "epoch": 15.840591290451458,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 39649
    },
    {
      "epoch": 15.840990811026767,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 39650
    },
    {
      "epoch": 15.841390331602078,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 39651
    },
    {
      "epoch": 15.841789852177387,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 39652
    },
    {
      "epoch": 15.842189372752697,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 39653
    },
    {
      "epoch": 15.842588893328006,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 39654
    },
    {
      "epoch": 15.842988413903315,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 39655
    },
    {
      "epoch": 15.843387934478626,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 39656
    },
    {
      "epoch": 15.843787455053935,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 39657
    },
    {
      "epoch": 15.844186975629245,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 39658
    },
    {
      "epoch": 15.844586496204554,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 39659
    },
    {
      "epoch": 15.844986016779863,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 39660
    },
    {
      "epoch": 15.845385537355174,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 39661
    },
    {
      "epoch": 15.845785057930483,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 39662
    },
    {
      "epoch": 15.846184578505794,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 39663
    },
    {
      "epoch": 15.846584099081102,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 39664
    },
    {
      "epoch": 15.846983619656413,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 39665
    },
    {
      "epoch": 15.847383140231722,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 39666
    },
    {
      "epoch": 15.847782660807031,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 39667
    },
    {
      "epoch": 15.848182181382342,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 39668
    },
    {
      "epoch": 15.84858170195765,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 39669
    },
    {
      "epoch": 15.848981222532961,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 39670
    },
    {
      "epoch": 15.84938074310827,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 39671
    },
    {
      "epoch": 15.849780263683579,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 39672
    },
    {
      "epoch": 15.85017978425889,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 39673
    },
    {
      "epoch": 15.850579304834199,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 39674
    },
    {
      "epoch": 15.85097882540951,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 39675
    },
    {
      "epoch": 15.851378345984818,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 39676
    },
    {
      "epoch": 15.851777866560127,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 39677
    },
    {
      "epoch": 15.852177387135438,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 39678
    },
    {
      "epoch": 15.852576907710747,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 39679
    },
    {
      "epoch": 15.852976428286057,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 39680
    },
    {
      "epoch": 15.853375948861366,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 39681
    },
    {
      "epoch": 15.853775469436677,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 39682
    },
    {
      "epoch": 15.854174990011986,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 39683
    },
    {
      "epoch": 15.854574510587295,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 39684
    },
    {
      "epoch": 15.854974031162605,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 39685
    },
    {
      "epoch": 15.855373551737914,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 39686
    },
    {
      "epoch": 15.855773072313225,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 39687
    },
    {
      "epoch": 15.856172592888534,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0111,
      "step": 39688
    },
    {
      "epoch": 15.856572113463843,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 39689
    },
    {
      "epoch": 15.856971634039153,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 39690
    },
    {
      "epoch": 15.857371154614462,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 39691
    },
    {
      "epoch": 15.857770675189773,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 39692
    },
    {
      "epoch": 15.858170195765082,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 39693
    },
    {
      "epoch": 15.85856971634039,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 39694
    },
    {
      "epoch": 15.858969236915701,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 39695
    },
    {
      "epoch": 15.85936875749101,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 39696
    },
    {
      "epoch": 15.859768278066321,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 39697
    },
    {
      "epoch": 15.86016779864163,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 39698
    },
    {
      "epoch": 15.86056731921694,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 39699
    },
    {
      "epoch": 15.86096683979225,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 39700
    },
    {
      "epoch": 15.861366360367558,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 39701
    },
    {
      "epoch": 15.861765880942869,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 39702
    },
    {
      "epoch": 15.862165401518178,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 39703
    },
    {
      "epoch": 15.862564922093489,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 39704
    },
    {
      "epoch": 15.862964442668797,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 39705
    },
    {
      "epoch": 15.863363963244106,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 39706
    },
    {
      "epoch": 15.863763483819417,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0082,
      "step": 39707
    },
    {
      "epoch": 15.864163004394726,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 39708
    },
    {
      "epoch": 15.864562524970037,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 39709
    },
    {
      "epoch": 15.864962045545346,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 39710
    },
    {
      "epoch": 15.865361566120654,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 39711
    },
    {
      "epoch": 15.865761086695965,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 39712
    },
    {
      "epoch": 15.866160607271274,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 39713
    },
    {
      "epoch": 15.866560127846585,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 39714
    },
    {
      "epoch": 15.866959648421894,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 39715
    },
    {
      "epoch": 15.867359168997204,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 39716
    },
    {
      "epoch": 15.867758689572513,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 39717
    },
    {
      "epoch": 15.868158210147822,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 39718
    },
    {
      "epoch": 15.868557730723133,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 39719
    },
    {
      "epoch": 15.868957251298442,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 39720
    },
    {
      "epoch": 15.869356771873752,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 39721
    },
    {
      "epoch": 15.869756292449061,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 39722
    },
    {
      "epoch": 15.87015581302437,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 39723
    },
    {
      "epoch": 15.87055533359968,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 39724
    },
    {
      "epoch": 15.87095485417499,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 39725
    },
    {
      "epoch": 15.8713543747503,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 39726
    },
    {
      "epoch": 15.87175389532561,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 39727
    },
    {
      "epoch": 15.872153415900918,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 39728
    },
    {
      "epoch": 15.872552936476229,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 39729
    },
    {
      "epoch": 15.872952457051538,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 39730
    },
    {
      "epoch": 15.873351977626848,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 39731
    },
    {
      "epoch": 15.873751498202157,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 39732
    },
    {
      "epoch": 15.874151018777468,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 39733
    },
    {
      "epoch": 15.874550539352777,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 39734
    },
    {
      "epoch": 15.874950059928086,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 39735
    },
    {
      "epoch": 15.875349580503396,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 39736
    },
    {
      "epoch": 15.875749101078705,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 39737
    },
    {
      "epoch": 15.876148621654016,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 39738
    },
    {
      "epoch": 15.876548142229325,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 39739
    },
    {
      "epoch": 15.876947662804634,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 39740
    },
    {
      "epoch": 15.877347183379944,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 39741
    },
    {
      "epoch": 15.877746703955253,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 39742
    },
    {
      "epoch": 15.878146224530564,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 39743
    },
    {
      "epoch": 15.878545745105873,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 39744
    },
    {
      "epoch": 15.878945265681182,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 39745
    },
    {
      "epoch": 15.879344786256492,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 39746
    },
    {
      "epoch": 15.879744306831801,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 39747
    },
    {
      "epoch": 15.880143827407112,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 39748
    },
    {
      "epoch": 15.880543347982421,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 39749
    },
    {
      "epoch": 15.880942868557732,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 39750
    },
    {
      "epoch": 15.88134238913304,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 39751
    },
    {
      "epoch": 15.88174190970835,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 39752
    },
    {
      "epoch": 15.88214143028366,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 39753
    },
    {
      "epoch": 15.882540950858969,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 39754
    },
    {
      "epoch": 15.88294047143428,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 39755
    },
    {
      "epoch": 15.883339992009589,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 39756
    },
    {
      "epoch": 15.883739512584897,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 39757
    },
    {
      "epoch": 15.884139033160208,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 39758
    },
    {
      "epoch": 15.884538553735517,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 39759
    },
    {
      "epoch": 15.884938074310828,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 39760
    },
    {
      "epoch": 15.885337594886137,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 39761
    },
    {
      "epoch": 15.885737115461446,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 39762
    },
    {
      "epoch": 15.886136636036756,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 39763
    },
    {
      "epoch": 15.886536156612065,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 39764
    },
    {
      "epoch": 15.886935677187376,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 39765
    },
    {
      "epoch": 15.887335197762685,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 39766
    },
    {
      "epoch": 15.887734718337995,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 39767
    },
    {
      "epoch": 15.888134238913304,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 39768
    },
    {
      "epoch": 15.888533759488613,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 39769
    },
    {
      "epoch": 15.888933280063924,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 39770
    },
    {
      "epoch": 15.889332800639233,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0125,
      "step": 39771
    },
    {
      "epoch": 15.889732321214543,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 39772
    },
    {
      "epoch": 15.890131841789852,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 39773
    },
    {
      "epoch": 15.890531362365161,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 39774
    },
    {
      "epoch": 15.890930882940472,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 39775
    },
    {
      "epoch": 15.89133040351578,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 39776
    },
    {
      "epoch": 15.891729924091091,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 39777
    },
    {
      "epoch": 15.8921294446664,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 39778
    },
    {
      "epoch": 15.89252896524171,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 39779
    },
    {
      "epoch": 15.89292848581702,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 39780
    },
    {
      "epoch": 15.893328006392329,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 39781
    },
    {
      "epoch": 15.89372752696764,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 39782
    },
    {
      "epoch": 15.894127047542948,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 39783
    },
    {
      "epoch": 15.894526568118257,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 39784
    },
    {
      "epoch": 15.894926088693568,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 39785
    },
    {
      "epoch": 15.895325609268877,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 39786
    },
    {
      "epoch": 15.895725129844188,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 39787
    },
    {
      "epoch": 15.896124650419496,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 39788
    },
    {
      "epoch": 15.896524170994805,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 39789
    },
    {
      "epoch": 15.896923691570116,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 39790
    },
    {
      "epoch": 15.897323212145425,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 39791
    },
    {
      "epoch": 15.897722732720736,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 39792
    },
    {
      "epoch": 15.898122253296044,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 39793
    },
    {
      "epoch": 15.898521773871355,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 39794
    },
    {
      "epoch": 15.898921294446664,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 39795
    },
    {
      "epoch": 15.899320815021973,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 39796
    },
    {
      "epoch": 15.899720335597284,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 39797
    },
    {
      "epoch": 15.900119856172592,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 39798
    },
    {
      "epoch": 15.900519376747903,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 39799
    },
    {
      "epoch": 15.900918897323212,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 39800
    },
    {
      "epoch": 15.901318417898521,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 39801
    },
    {
      "epoch": 15.901717938473832,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 39802
    },
    {
      "epoch": 15.90211745904914,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 39803
    },
    {
      "epoch": 15.902516979624451,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 39804
    },
    {
      "epoch": 15.90291650019976,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 39805
    },
    {
      "epoch": 15.903316020775069,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 39806
    },
    {
      "epoch": 15.90371554135038,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 39807
    },
    {
      "epoch": 15.904115061925689,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 39808
    },
    {
      "epoch": 15.904514582501,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 39809
    },
    {
      "epoch": 15.904914103076308,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 39810
    },
    {
      "epoch": 15.905313623651619,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 39811
    },
    {
      "epoch": 15.905713144226928,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 39812
    },
    {
      "epoch": 15.906112664802237,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 39813
    },
    {
      "epoch": 15.906512185377547,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 39814
    },
    {
      "epoch": 15.906911705952856,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 39815
    },
    {
      "epoch": 15.907311226528167,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 39816
    },
    {
      "epoch": 15.907710747103476,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 39817
    },
    {
      "epoch": 15.908110267678785,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 39818
    },
    {
      "epoch": 15.908509788254095,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 39819
    },
    {
      "epoch": 15.908909308829404,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 39820
    },
    {
      "epoch": 15.909308829404715,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 39821
    },
    {
      "epoch": 15.909708349980024,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 39822
    },
    {
      "epoch": 15.910107870555333,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 39823
    },
    {
      "epoch": 15.910507391130643,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 39824
    },
    {
      "epoch": 15.910906911705952,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 39825
    },
    {
      "epoch": 15.911306432281263,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 39826
    },
    {
      "epoch": 15.911705952856572,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 39827
    },
    {
      "epoch": 15.912105473431883,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 39828
    },
    {
      "epoch": 15.912504994007191,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 39829
    },
    {
      "epoch": 15.9129045145825,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 39830
    },
    {
      "epoch": 15.913304035157811,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 39831
    },
    {
      "epoch": 15.91370355573312,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 39832
    },
    {
      "epoch": 15.91410307630843,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 39833
    },
    {
      "epoch": 15.91450259688374,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 39834
    },
    {
      "epoch": 15.914902117459048,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 39835
    },
    {
      "epoch": 15.915301638034359,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 39836
    },
    {
      "epoch": 15.915701158609668,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 39837
    },
    {
      "epoch": 15.916100679184979,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 39838
    },
    {
      "epoch": 15.916500199760288,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 39839
    },
    {
      "epoch": 15.916899720335596,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 39840
    },
    {
      "epoch": 15.917299240910907,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 39841
    },
    {
      "epoch": 15.917698761486216,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 39842
    },
    {
      "epoch": 15.918098282061527,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 39843
    },
    {
      "epoch": 15.918497802636836,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 39844
    },
    {
      "epoch": 15.918897323212146,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 39845
    },
    {
      "epoch": 15.919296843787455,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 39846
    },
    {
      "epoch": 15.919696364362764,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 39847
    },
    {
      "epoch": 15.920095884938075,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 39848
    },
    {
      "epoch": 15.920495405513384,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 39849
    },
    {
      "epoch": 15.920894926088694,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 39850
    },
    {
      "epoch": 15.921294446664003,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 39851
    },
    {
      "epoch": 15.921693967239312,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 39852
    },
    {
      "epoch": 15.922093487814623,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 39853
    },
    {
      "epoch": 15.922493008389932,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 39854
    },
    {
      "epoch": 15.922892528965242,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 39855
    },
    {
      "epoch": 15.923292049540551,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 39856
    },
    {
      "epoch": 15.92369157011586,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 39857
    },
    {
      "epoch": 15.92409109069117,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 39858
    },
    {
      "epoch": 15.92449061126648,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 39859
    },
    {
      "epoch": 15.92489013184179,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 39860
    },
    {
      "epoch": 15.9252896524171,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 39861
    },
    {
      "epoch": 15.92568917299241,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0077,
      "step": 39862
    },
    {
      "epoch": 15.926088693567719,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 39863
    },
    {
      "epoch": 15.926488214143028,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 39864
    },
    {
      "epoch": 15.926887734718338,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0124,
      "step": 39865
    },
    {
      "epoch": 15.927287255293647,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 39866
    },
    {
      "epoch": 15.927686775868958,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 39867
    },
    {
      "epoch": 15.928086296444267,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 39868
    },
    {
      "epoch": 15.928485817019576,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 39869
    },
    {
      "epoch": 15.928885337594886,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 39870
    },
    {
      "epoch": 15.929284858170195,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 39871
    },
    {
      "epoch": 15.929684378745506,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 39872
    },
    {
      "epoch": 15.930083899320815,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 39873
    },
    {
      "epoch": 15.930483419896124,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 39874
    },
    {
      "epoch": 15.930882940471434,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 39875
    },
    {
      "epoch": 15.931282461046743,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 39876
    },
    {
      "epoch": 15.931681981622054,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 39877
    },
    {
      "epoch": 15.932081502197363,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 39878
    },
    {
      "epoch": 15.932481022772674,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 39879
    },
    {
      "epoch": 15.932880543347983,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 39880
    },
    {
      "epoch": 15.933280063923291,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 39881
    },
    {
      "epoch": 15.933679584498602,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 39882
    },
    {
      "epoch": 15.934079105073911,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 39883
    },
    {
      "epoch": 15.934478625649222,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 39884
    },
    {
      "epoch": 15.93487814622453,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 39885
    },
    {
      "epoch": 15.93527766679984,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 39886
    },
    {
      "epoch": 15.93567718737515,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 39887
    },
    {
      "epoch": 15.936076707950459,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 39888
    },
    {
      "epoch": 15.93647622852577,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 39889
    },
    {
      "epoch": 15.936875749101079,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 39890
    },
    {
      "epoch": 15.937275269676388,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 39891
    },
    {
      "epoch": 15.937674790251698,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 39892
    },
    {
      "epoch": 15.938074310827007,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 39893
    },
    {
      "epoch": 15.938473831402318,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 39894
    },
    {
      "epoch": 15.938873351977627,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 39895
    },
    {
      "epoch": 15.939272872552937,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 39896
    },
    {
      "epoch": 15.939672393128246,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 39897
    },
    {
      "epoch": 15.940071913703555,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 39898
    },
    {
      "epoch": 15.940471434278866,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 39899
    },
    {
      "epoch": 15.940870954854175,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 39900
    },
    {
      "epoch": 15.941270475429485,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 39901
    },
    {
      "epoch": 15.941669996004794,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 39902
    },
    {
      "epoch": 15.942069516580103,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 39903
    },
    {
      "epoch": 15.942469037155414,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 39904
    },
    {
      "epoch": 15.942868557730723,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 39905
    },
    {
      "epoch": 15.943268078306033,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 39906
    },
    {
      "epoch": 15.943667598881342,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 39907
    },
    {
      "epoch": 15.944067119456651,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 39908
    },
    {
      "epoch": 15.944466640031962,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 39909
    },
    {
      "epoch": 15.94486616060727,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 39910
    },
    {
      "epoch": 15.945265681182581,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 39911
    },
    {
      "epoch": 15.94566520175789,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 39912
    },
    {
      "epoch": 15.946064722333201,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 39913
    },
    {
      "epoch": 15.94646424290851,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 39914
    },
    {
      "epoch": 15.946863763483819,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 39915
    },
    {
      "epoch": 15.94726328405913,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 39916
    },
    {
      "epoch": 15.947662804634438,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 39917
    },
    {
      "epoch": 15.948062325209749,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 39918
    },
    {
      "epoch": 15.948461845785058,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 39919
    },
    {
      "epoch": 15.948861366360367,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 39920
    },
    {
      "epoch": 15.949260886935678,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 39921
    },
    {
      "epoch": 15.949660407510986,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 39922
    },
    {
      "epoch": 15.950059928086297,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 39923
    },
    {
      "epoch": 15.950459448661606,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 39924
    },
    {
      "epoch": 15.950858969236915,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0121,
      "step": 39925
    },
    {
      "epoch": 15.951258489812226,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 39926
    },
    {
      "epoch": 15.951658010387535,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 39927
    },
    {
      "epoch": 15.952057530962845,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 39928
    },
    {
      "epoch": 15.952457051538154,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 39929
    },
    {
      "epoch": 15.952856572113465,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 39930
    },
    {
      "epoch": 15.953256092688774,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 39931
    },
    {
      "epoch": 15.953655613264083,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 39932
    },
    {
      "epoch": 15.954055133839393,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 39933
    },
    {
      "epoch": 15.954454654414702,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 39934
    },
    {
      "epoch": 15.954854174990013,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 39935
    },
    {
      "epoch": 15.955253695565322,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 39936
    },
    {
      "epoch": 15.95565321614063,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 39937
    },
    {
      "epoch": 15.956052736715941,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 39938
    },
    {
      "epoch": 15.95645225729125,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 39939
    },
    {
      "epoch": 15.95685177786656,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 39940
    },
    {
      "epoch": 15.95725129844187,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 39941
    },
    {
      "epoch": 15.957650819017179,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 39942
    },
    {
      "epoch": 15.95805033959249,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 39943
    },
    {
      "epoch": 15.958449860167798,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 39944
    },
    {
      "epoch": 15.958849380743109,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 39945
    },
    {
      "epoch": 15.959248901318418,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 39946
    },
    {
      "epoch": 15.959648421893728,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 39947
    },
    {
      "epoch": 15.960047942469037,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 39948
    },
    {
      "epoch": 15.960447463044346,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 39949
    },
    {
      "epoch": 15.960846983619657,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 39950
    },
    {
      "epoch": 15.961246504194966,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 39951
    },
    {
      "epoch": 15.961646024770276,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 39952
    },
    {
      "epoch": 15.962045545345585,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 39953
    },
    {
      "epoch": 15.962445065920894,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 39954
    },
    {
      "epoch": 15.962844586496205,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 39955
    },
    {
      "epoch": 15.963244107071514,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 39956
    },
    {
      "epoch": 15.963643627646825,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 39957
    },
    {
      "epoch": 15.964043148222133,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 39958
    },
    {
      "epoch": 15.964442668797442,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 39959
    },
    {
      "epoch": 15.964842189372753,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 39960
    },
    {
      "epoch": 15.965241709948062,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 39961
    },
    {
      "epoch": 15.965641230523373,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 39962
    },
    {
      "epoch": 15.966040751098681,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 39963
    },
    {
      "epoch": 15.96644027167399,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 39964
    },
    {
      "epoch": 15.966839792249301,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 39965
    },
    {
      "epoch": 15.96723931282461,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 39966
    },
    {
      "epoch": 15.96763883339992,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 39967
    },
    {
      "epoch": 15.96803835397523,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 39968
    },
    {
      "epoch": 15.968437874550538,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 39969
    },
    {
      "epoch": 15.968837395125849,
      "grad_norm": 0.036376953125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 39970
    },
    {
      "epoch": 15.969236915701158,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0466,
      "step": 39971
    },
    {
      "epoch": 15.969636436276469,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 39972
    },
    {
      "epoch": 15.970035956851778,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 39973
    },
    {
      "epoch": 15.970435477427088,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 39974
    },
    {
      "epoch": 15.970834998002397,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 39975
    },
    {
      "epoch": 15.971234518577706,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 39976
    },
    {
      "epoch": 15.971634039153017,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 39977
    },
    {
      "epoch": 15.972033559728326,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 39978
    },
    {
      "epoch": 15.972433080303636,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 39979
    },
    {
      "epoch": 15.972832600878945,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 39980
    },
    {
      "epoch": 15.973232121454254,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 39981
    },
    {
      "epoch": 15.973631642029565,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 39982
    },
    {
      "epoch": 15.974031162604874,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 39983
    },
    {
      "epoch": 15.974430683180184,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 39984
    },
    {
      "epoch": 15.974830203755493,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 39985
    },
    {
      "epoch": 15.975229724330802,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 39986
    },
    {
      "epoch": 15.975629244906113,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 39987
    },
    {
      "epoch": 15.976028765481422,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 39988
    },
    {
      "epoch": 15.976428286056732,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 39989
    },
    {
      "epoch": 15.976827806632041,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 39990
    },
    {
      "epoch": 15.977227327207352,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0081,
      "step": 39991
    },
    {
      "epoch": 15.97762684778266,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 39992
    },
    {
      "epoch": 15.97802636835797,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 39993
    },
    {
      "epoch": 15.97842588893328,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 39994
    },
    {
      "epoch": 15.97882540950859,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 39995
    },
    {
      "epoch": 15.9792249300839,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 39996
    },
    {
      "epoch": 15.979624450659209,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 39997
    },
    {
      "epoch": 15.980023971234518,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0106,
      "step": 39998
    },
    {
      "epoch": 15.980423491809828,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 39999
    },
    {
      "epoch": 15.980823012385137,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0446,
      "step": 40000
    },
    {
      "epoch": 15.981222532960448,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 40001
    },
    {
      "epoch": 15.981622053535757,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 40002
    },
    {
      "epoch": 15.982021574111066,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 40003
    },
    {
      "epoch": 15.982421094686377,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 40004
    },
    {
      "epoch": 15.982820615261685,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 40005
    },
    {
      "epoch": 15.983220135836996,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 40006
    },
    {
      "epoch": 15.983619656412305,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 40007
    },
    {
      "epoch": 15.984019176987616,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 40008
    },
    {
      "epoch": 15.984418697562925,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 40009
    },
    {
      "epoch": 15.984818218138233,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 40010
    },
    {
      "epoch": 15.985217738713544,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 40011
    },
    {
      "epoch": 15.985617259288853,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0135,
      "step": 40012
    },
    {
      "epoch": 15.986016779864164,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 40013
    },
    {
      "epoch": 15.986416300439473,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 40014
    },
    {
      "epoch": 15.986815821014781,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 40015
    },
    {
      "epoch": 15.987215341590092,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 40016
    },
    {
      "epoch": 15.987614862165401,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 40017
    },
    {
      "epoch": 15.988014382740712,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 40018
    },
    {
      "epoch": 15.98841390331602,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 40019
    },
    {
      "epoch": 15.98881342389133,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 40020
    },
    {
      "epoch": 15.98921294446664,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 40021
    },
    {
      "epoch": 15.989612465041949,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 40022
    },
    {
      "epoch": 15.99001198561726,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 40023
    },
    {
      "epoch": 15.990411506192569,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 40024
    },
    {
      "epoch": 15.99081102676788,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 40025
    },
    {
      "epoch": 15.991210547343188,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 40026
    },
    {
      "epoch": 15.991610067918497,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0124,
      "step": 40027
    },
    {
      "epoch": 15.992009588493808,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 40028
    },
    {
      "epoch": 15.992409109069117,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 40029
    },
    {
      "epoch": 15.992808629644427,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 40030
    },
    {
      "epoch": 15.993208150219736,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 40031
    },
    {
      "epoch": 15.993607670795045,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 40032
    },
    {
      "epoch": 15.994007191370356,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 40033
    },
    {
      "epoch": 15.994406711945665,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 40034
    },
    {
      "epoch": 15.994806232520975,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 40035
    },
    {
      "epoch": 15.995205753096284,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 40036
    },
    {
      "epoch": 15.995605273671593,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 40037
    },
    {
      "epoch": 15.996004794246904,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 40038
    },
    {
      "epoch": 15.996404314822213,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 40039
    },
    {
      "epoch": 15.996803835397523,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 40040
    },
    {
      "epoch": 15.997203355972832,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 40041
    },
    {
      "epoch": 15.997602876548143,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 40042
    },
    {
      "epoch": 15.998002397123452,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 40043
    },
    {
      "epoch": 15.99840191769876,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 40044
    },
    {
      "epoch": 15.998801438274072,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 40045
    },
    {
      "epoch": 15.99920095884938,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 40046
    },
    {
      "epoch": 15.999600479424691,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 40047
    },
    {
      "epoch": 16.0,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 40048
    },
    {
      "epoch": 16.00039952057531,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 40049
    },
    {
      "epoch": 16.000799041150618,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 40050
    },
    {
      "epoch": 16.00119856172593,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 40051
    },
    {
      "epoch": 16.00159808230124,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 40052
    },
    {
      "epoch": 16.00199760287655,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 40053
    },
    {
      "epoch": 16.002397123451857,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 40054
    },
    {
      "epoch": 16.002796644027168,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 40055
    },
    {
      "epoch": 16.00319616460248,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 40056
    },
    {
      "epoch": 16.003595685177785,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 40057
    },
    {
      "epoch": 16.003995205753096,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 40058
    },
    {
      "epoch": 16.004394726328407,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 40059
    },
    {
      "epoch": 16.004794246903714,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 40060
    },
    {
      "epoch": 16.005193767479025,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 40061
    },
    {
      "epoch": 16.005593288054335,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 40062
    },
    {
      "epoch": 16.005992808629646,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 40063
    },
    {
      "epoch": 16.006392329204953,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 40064
    },
    {
      "epoch": 16.006791849780264,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 40065
    },
    {
      "epoch": 16.007191370355574,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 40066
    },
    {
      "epoch": 16.00759089093088,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 40067
    },
    {
      "epoch": 16.007990411506192,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 40068
    },
    {
      "epoch": 16.008389932081503,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 40069
    },
    {
      "epoch": 16.008789452656814,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 40070
    },
    {
      "epoch": 16.00918897323212,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 40071
    },
    {
      "epoch": 16.00958849380743,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 40072
    },
    {
      "epoch": 16.009988014382742,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 40073
    },
    {
      "epoch": 16.01038753495805,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 40074
    },
    {
      "epoch": 16.01078705553336,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 40075
    },
    {
      "epoch": 16.01118657610867,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 40076
    },
    {
      "epoch": 16.011586096683978,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 40077
    },
    {
      "epoch": 16.01198561725929,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 40078
    },
    {
      "epoch": 16.0123851378346,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 40079
    },
    {
      "epoch": 16.01278465840991,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 40080
    },
    {
      "epoch": 16.013184178985217,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 40081
    },
    {
      "epoch": 16.013583699560527,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 40082
    },
    {
      "epoch": 16.013983220135838,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 40083
    },
    {
      "epoch": 16.014382740711145,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 40084
    },
    {
      "epoch": 16.014782261286456,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 40085
    },
    {
      "epoch": 16.015181781861767,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 40086
    },
    {
      "epoch": 16.015581302437077,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 40087
    },
    {
      "epoch": 16.015980823012384,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 40088
    },
    {
      "epoch": 16.016380343587695,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 40089
    },
    {
      "epoch": 16.016779864163006,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 40090
    },
    {
      "epoch": 16.017179384738313,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 40091
    },
    {
      "epoch": 16.017578905313623,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 40092
    },
    {
      "epoch": 16.017978425888934,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 40093
    },
    {
      "epoch": 16.01837794646424,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 40094
    },
    {
      "epoch": 16.018777467039552,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 40095
    },
    {
      "epoch": 16.019176987614863,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 40096
    },
    {
      "epoch": 16.019576508190173,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 40097
    },
    {
      "epoch": 16.01997602876548,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 40098
    },
    {
      "epoch": 16.02037554934079,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 40099
    },
    {
      "epoch": 16.020775069916102,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 40100
    },
    {
      "epoch": 16.02117459049141,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 40101
    },
    {
      "epoch": 16.02157411106672,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 40102
    },
    {
      "epoch": 16.02197363164203,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 40103
    },
    {
      "epoch": 16.02237315221734,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 40104
    },
    {
      "epoch": 16.022772672792648,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 40105
    },
    {
      "epoch": 16.02317219336796,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 40106
    },
    {
      "epoch": 16.02357171394327,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 40107
    },
    {
      "epoch": 16.023971234518577,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0083,
      "step": 40108
    },
    {
      "epoch": 16.024370755093887,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 40109
    },
    {
      "epoch": 16.024770275669198,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 40110
    },
    {
      "epoch": 16.025169796244505,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 40111
    },
    {
      "epoch": 16.025569316819816,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 40112
    },
    {
      "epoch": 16.025968837395126,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 40113
    },
    {
      "epoch": 16.026368357970437,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 40114
    },
    {
      "epoch": 16.026767878545744,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 40115
    },
    {
      "epoch": 16.027167399121055,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 40116
    },
    {
      "epoch": 16.027566919696365,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 40117
    },
    {
      "epoch": 16.027966440271673,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 40118
    },
    {
      "epoch": 16.028365960846983,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 40119
    },
    {
      "epoch": 16.028765481422294,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 40120
    },
    {
      "epoch": 16.029165001997605,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0119,
      "step": 40121
    },
    {
      "epoch": 16.02956452257291,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 40122
    },
    {
      "epoch": 16.029964043148222,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 40123
    },
    {
      "epoch": 16.030363563723533,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 40124
    },
    {
      "epoch": 16.03076308429884,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 40125
    },
    {
      "epoch": 16.03116260487415,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 40126
    },
    {
      "epoch": 16.03156212544946,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 40127
    },
    {
      "epoch": 16.03196164602477,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 40128
    },
    {
      "epoch": 16.03236116660008,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 40129
    },
    {
      "epoch": 16.03276068717539,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 40130
    },
    {
      "epoch": 16.0331602077507,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 40131
    },
    {
      "epoch": 16.033559728326008,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 40132
    },
    {
      "epoch": 16.03395924890132,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 40133
    },
    {
      "epoch": 16.03435876947663,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 40134
    },
    {
      "epoch": 16.034758290051936,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 40135
    },
    {
      "epoch": 16.035157810627247,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 40136
    },
    {
      "epoch": 16.035557331202558,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 40137
    },
    {
      "epoch": 16.03595685177787,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 40138
    },
    {
      "epoch": 16.036356372353175,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 40139
    },
    {
      "epoch": 16.036755892928486,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 40140
    },
    {
      "epoch": 16.037155413503797,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 40141
    },
    {
      "epoch": 16.037554934079104,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 40142
    },
    {
      "epoch": 16.037954454654415,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 40143
    },
    {
      "epoch": 16.038353975229725,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 40144
    },
    {
      "epoch": 16.038753495805032,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 40145
    },
    {
      "epoch": 16.039153016380343,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 40146
    },
    {
      "epoch": 16.039552536955654,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 40147
    },
    {
      "epoch": 16.039952057530964,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 40148
    },
    {
      "epoch": 16.04035157810627,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 40149
    },
    {
      "epoch": 16.040751098681582,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 40150
    },
    {
      "epoch": 16.041150619256893,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 40151
    },
    {
      "epoch": 16.0415501398322,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 40152
    },
    {
      "epoch": 16.04194966040751,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 40153
    },
    {
      "epoch": 16.04234918098282,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 40154
    },
    {
      "epoch": 16.04274870155813,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 40155
    },
    {
      "epoch": 16.04314822213344,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 40156
    },
    {
      "epoch": 16.04354774270875,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 40157
    },
    {
      "epoch": 16.04394726328406,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 40158
    },
    {
      "epoch": 16.044346783859368,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 40159
    },
    {
      "epoch": 16.04474630443468,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.0058,
      "step": 40160
    },
    {
      "epoch": 16.04514582500999,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 40161
    },
    {
      "epoch": 16.045545345585296,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 40162
    },
    {
      "epoch": 16.045944866160607,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 40163
    },
    {
      "epoch": 16.046344386735917,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 40164
    },
    {
      "epoch": 16.046743907311228,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 40165
    },
    {
      "epoch": 16.047143427886535,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 40166
    },
    {
      "epoch": 16.047542948461846,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 40167
    },
    {
      "epoch": 16.047942469037157,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 40168
    },
    {
      "epoch": 16.048341989612464,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 40169
    },
    {
      "epoch": 16.048741510187774,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.05,
      "step": 40170
    },
    {
      "epoch": 16.049141030763085,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 40171
    },
    {
      "epoch": 16.049540551338392,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 40172
    },
    {
      "epoch": 16.049940071913703,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 40173
    },
    {
      "epoch": 16.050339592489014,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 40174
    },
    {
      "epoch": 16.050739113064324,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0112,
      "step": 40175
    },
    {
      "epoch": 16.05113863363963,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 40176
    },
    {
      "epoch": 16.051538154214942,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 40177
    },
    {
      "epoch": 16.051937674790253,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 40178
    },
    {
      "epoch": 16.05233719536556,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 40179
    },
    {
      "epoch": 16.05273671594087,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 40180
    },
    {
      "epoch": 16.05313623651618,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 40181
    },
    {
      "epoch": 16.053535757091492,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 40182
    },
    {
      "epoch": 16.0539352776668,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 40183
    },
    {
      "epoch": 16.05433479824211,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 40184
    },
    {
      "epoch": 16.05473431881742,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 40185
    },
    {
      "epoch": 16.055133839392727,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 40186
    },
    {
      "epoch": 16.055533359968038,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 40187
    },
    {
      "epoch": 16.05593288054335,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 40188
    },
    {
      "epoch": 16.056332401118656,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 40189
    },
    {
      "epoch": 16.056731921693967,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 40190
    },
    {
      "epoch": 16.057131442269277,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 40191
    },
    {
      "epoch": 16.057530962844588,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 40192
    },
    {
      "epoch": 16.057930483419895,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 40193
    },
    {
      "epoch": 16.058330003995206,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 40194
    },
    {
      "epoch": 16.058729524570516,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 40195
    },
    {
      "epoch": 16.059129045145824,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 40196
    },
    {
      "epoch": 16.059528565721134,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 40197
    },
    {
      "epoch": 16.059928086296445,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 40198
    },
    {
      "epoch": 16.060327606871756,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 40199
    },
    {
      "epoch": 16.060727127447063,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 40200
    },
    {
      "epoch": 16.061126648022373,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 40201
    },
    {
      "epoch": 16.061526168597684,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 40202
    },
    {
      "epoch": 16.06192568917299,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 40203
    },
    {
      "epoch": 16.062325209748302,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 40204
    },
    {
      "epoch": 16.062724730323612,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 40205
    },
    {
      "epoch": 16.06312425089892,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 40206
    },
    {
      "epoch": 16.06352377147423,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0131,
      "step": 40207
    },
    {
      "epoch": 16.06392329204954,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 40208
    },
    {
      "epoch": 16.06432281262485,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 40209
    },
    {
      "epoch": 16.06472233320016,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 40210
    },
    {
      "epoch": 16.06512185377547,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 40211
    },
    {
      "epoch": 16.06552137435078,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 40212
    },
    {
      "epoch": 16.065920894926087,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 40213
    },
    {
      "epoch": 16.066320415501398,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 40214
    },
    {
      "epoch": 16.06671993607671,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 40215
    },
    {
      "epoch": 16.06711945665202,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 40216
    },
    {
      "epoch": 16.067518977227326,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 40217
    },
    {
      "epoch": 16.067918497802637,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 40218
    },
    {
      "epoch": 16.068318018377948,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 40219
    },
    {
      "epoch": 16.068717538953255,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 40220
    },
    {
      "epoch": 16.069117059528566,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 40221
    },
    {
      "epoch": 16.069516580103876,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 40222
    },
    {
      "epoch": 16.069916100679183,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 40223
    },
    {
      "epoch": 16.070315621254494,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 40224
    },
    {
      "epoch": 16.070715141829805,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 40225
    },
    {
      "epoch": 16.071114662405115,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 40226
    },
    {
      "epoch": 16.071514182980422,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 40227
    },
    {
      "epoch": 16.071913703555733,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0489,
      "step": 40228
    },
    {
      "epoch": 16.072313224131044,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 40229
    },
    {
      "epoch": 16.07271274470635,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 40230
    },
    {
      "epoch": 16.07311226528166,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 40231
    },
    {
      "epoch": 16.073511785856972,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0032,
      "step": 40232
    },
    {
      "epoch": 16.073911306432283,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 40233
    },
    {
      "epoch": 16.07431082700759,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 40234
    },
    {
      "epoch": 16.0747103475829,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 40235
    },
    {
      "epoch": 16.07510986815821,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 40236
    },
    {
      "epoch": 16.07550938873352,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 40237
    },
    {
      "epoch": 16.07590890930883,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 40238
    },
    {
      "epoch": 16.07630842988414,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 40239
    },
    {
      "epoch": 16.076707950459447,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 40240
    },
    {
      "epoch": 16.077107471034758,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0071,
      "step": 40241
    },
    {
      "epoch": 16.07750699161007,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 40242
    },
    {
      "epoch": 16.07790651218538,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 40243
    },
    {
      "epoch": 16.078306032760686,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 40244
    },
    {
      "epoch": 16.078705553335997,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 40245
    },
    {
      "epoch": 16.079105073911307,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 40246
    },
    {
      "epoch": 16.079504594486615,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 40247
    },
    {
      "epoch": 16.079904115061925,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 40248
    },
    {
      "epoch": 16.080303635637236,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 40249
    },
    {
      "epoch": 16.080703156212547,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 40250
    },
    {
      "epoch": 16.081102676787854,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 40251
    },
    {
      "epoch": 16.081502197363164,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 40252
    },
    {
      "epoch": 16.081901717938475,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 40253
    },
    {
      "epoch": 16.082301238513782,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 40254
    },
    {
      "epoch": 16.082700759089093,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 40255
    },
    {
      "epoch": 16.083100279664404,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 40256
    },
    {
      "epoch": 16.08349980023971,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 40257
    },
    {
      "epoch": 16.08389932081502,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 40258
    },
    {
      "epoch": 16.084298841390332,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 40259
    },
    {
      "epoch": 16.084698361965643,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 40260
    },
    {
      "epoch": 16.08509788254095,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 40261
    },
    {
      "epoch": 16.08549740311626,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 40262
    },
    {
      "epoch": 16.08589692369157,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 40263
    },
    {
      "epoch": 16.08629644426688,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 40264
    },
    {
      "epoch": 16.08669596484219,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 40265
    },
    {
      "epoch": 16.0870954854175,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 40266
    },
    {
      "epoch": 16.08749500599281,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 40267
    },
    {
      "epoch": 16.087894526568117,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 40268
    },
    {
      "epoch": 16.088294047143428,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 40269
    },
    {
      "epoch": 16.08869356771874,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 40270
    },
    {
      "epoch": 16.089093088294046,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 40271
    },
    {
      "epoch": 16.089492608869357,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 40272
    },
    {
      "epoch": 16.089892129444667,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 40273
    },
    {
      "epoch": 16.090291650019974,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 40274
    },
    {
      "epoch": 16.090691170595285,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 40275
    },
    {
      "epoch": 16.091090691170596,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 40276
    },
    {
      "epoch": 16.091490211745906,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 40277
    },
    {
      "epoch": 16.091889732321214,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 40278
    },
    {
      "epoch": 16.092289252896524,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 40279
    },
    {
      "epoch": 16.092688773471835,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 40280
    },
    {
      "epoch": 16.093088294047142,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 40281
    },
    {
      "epoch": 16.093487814622453,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 40282
    },
    {
      "epoch": 16.093887335197763,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 40283
    },
    {
      "epoch": 16.094286855773074,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 40284
    },
    {
      "epoch": 16.09468637634838,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 40285
    },
    {
      "epoch": 16.095085896923692,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 40286
    },
    {
      "epoch": 16.095485417499003,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 40287
    },
    {
      "epoch": 16.09588493807431,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 40288
    },
    {
      "epoch": 16.09628445864962,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 40289
    },
    {
      "epoch": 16.09668397922493,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 40290
    },
    {
      "epoch": 16.097083499800238,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 40291
    },
    {
      "epoch": 16.09748302037555,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 40292
    },
    {
      "epoch": 16.09788254095086,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 40293
    },
    {
      "epoch": 16.09828206152617,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 40294
    },
    {
      "epoch": 16.098681582101477,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 40295
    },
    {
      "epoch": 16.099081102676788,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 40296
    },
    {
      "epoch": 16.0994806232521,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 40297
    },
    {
      "epoch": 16.099880143827406,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 40298
    },
    {
      "epoch": 16.100279664402716,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 40299
    },
    {
      "epoch": 16.100679184978027,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 40300
    },
    {
      "epoch": 16.101078705553338,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 40301
    },
    {
      "epoch": 16.101478226128645,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 40302
    },
    {
      "epoch": 16.101877746703956,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 40303
    },
    {
      "epoch": 16.102277267279266,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 40304
    },
    {
      "epoch": 16.102676787854573,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 40305
    },
    {
      "epoch": 16.103076308429884,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 40306
    },
    {
      "epoch": 16.103475829005195,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 40307
    },
    {
      "epoch": 16.103875349580502,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 40308
    },
    {
      "epoch": 16.104274870155812,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 40309
    },
    {
      "epoch": 16.104674390731123,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 40310
    },
    {
      "epoch": 16.105073911306434,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 40311
    },
    {
      "epoch": 16.10547343188174,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 40312
    },
    {
      "epoch": 16.10587295245705,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 40313
    },
    {
      "epoch": 16.106272473032362,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 40314
    },
    {
      "epoch": 16.10667199360767,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 40315
    },
    {
      "epoch": 16.10707151418298,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 40316
    },
    {
      "epoch": 16.10747103475829,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 40317
    },
    {
      "epoch": 16.107870555333598,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 40318
    },
    {
      "epoch": 16.10827007590891,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 40319
    },
    {
      "epoch": 16.10866959648422,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 40320
    },
    {
      "epoch": 16.10906911705953,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 40321
    },
    {
      "epoch": 16.109468637634837,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 40322
    },
    {
      "epoch": 16.109868158210148,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 40323
    },
    {
      "epoch": 16.11026767878546,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 40324
    },
    {
      "epoch": 16.110667199360766,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 40325
    },
    {
      "epoch": 16.111066719936076,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 40326
    },
    {
      "epoch": 16.111466240511387,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 40327
    },
    {
      "epoch": 16.111865761086698,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 40328
    },
    {
      "epoch": 16.112265281662005,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 40329
    },
    {
      "epoch": 16.112664802237315,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 40330
    },
    {
      "epoch": 16.113064322812626,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 40331
    },
    {
      "epoch": 16.113463843387933,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 40332
    },
    {
      "epoch": 16.113863363963244,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 40333
    },
    {
      "epoch": 16.114262884538554,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 40334
    },
    {
      "epoch": 16.11466240511386,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 40335
    },
    {
      "epoch": 16.115061925689172,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 40336
    },
    {
      "epoch": 16.115461446264483,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 40337
    },
    {
      "epoch": 16.115860966839794,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 40338
    },
    {
      "epoch": 16.1162604874151,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 40339
    },
    {
      "epoch": 16.11666000799041,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 40340
    },
    {
      "epoch": 16.117059528565722,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 40341
    },
    {
      "epoch": 16.11745904914103,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 40342
    },
    {
      "epoch": 16.11785856971634,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 40343
    },
    {
      "epoch": 16.11825809029165,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 40344
    },
    {
      "epoch": 16.11865761086696,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 40345
    },
    {
      "epoch": 16.11905713144227,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 40346
    },
    {
      "epoch": 16.11945665201758,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 40347
    },
    {
      "epoch": 16.11985617259289,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 40348
    },
    {
      "epoch": 16.120255693168197,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 40349
    },
    {
      "epoch": 16.120655213743508,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 40350
    },
    {
      "epoch": 16.121054734318818,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 40351
    },
    {
      "epoch": 16.121454254894125,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 40352
    },
    {
      "epoch": 16.121853775469436,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 40353
    },
    {
      "epoch": 16.122253296044747,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 40354
    },
    {
      "epoch": 16.122652816620057,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 40355
    },
    {
      "epoch": 16.123052337195364,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 40356
    },
    {
      "epoch": 16.123451857770675,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 40357
    },
    {
      "epoch": 16.123851378345986,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 40358
    },
    {
      "epoch": 16.124250898921293,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 40359
    },
    {
      "epoch": 16.124650419496604,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 40360
    },
    {
      "epoch": 16.125049940071914,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 40361
    },
    {
      "epoch": 16.125449460647225,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 40362
    },
    {
      "epoch": 16.125848981222532,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 40363
    },
    {
      "epoch": 16.126248501797843,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 40364
    },
    {
      "epoch": 16.126648022373153,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 40365
    },
    {
      "epoch": 16.12704754294846,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 40366
    },
    {
      "epoch": 16.12744706352377,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 40367
    },
    {
      "epoch": 16.127846584099082,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 40368
    },
    {
      "epoch": 16.12824610467439,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 40369
    },
    {
      "epoch": 16.1286456252497,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 40370
    },
    {
      "epoch": 16.12904514582501,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 40371
    },
    {
      "epoch": 16.12944466640032,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 40372
    },
    {
      "epoch": 16.129844186975628,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 40373
    },
    {
      "epoch": 16.13024370755094,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 40374
    },
    {
      "epoch": 16.13064322812625,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 40375
    },
    {
      "epoch": 16.131042748701557,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 40376
    },
    {
      "epoch": 16.131442269276867,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 40377
    },
    {
      "epoch": 16.131841789852178,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 40378
    },
    {
      "epoch": 16.13224131042749,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 40379
    },
    {
      "epoch": 16.132640831002796,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 40380
    },
    {
      "epoch": 16.133040351578106,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 40381
    },
    {
      "epoch": 16.133439872153417,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 40382
    },
    {
      "epoch": 16.133839392728724,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 40383
    },
    {
      "epoch": 16.134238913304035,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 40384
    },
    {
      "epoch": 16.134638433879346,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 40385
    },
    {
      "epoch": 16.135037954454653,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 40386
    },
    {
      "epoch": 16.135437475029963,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 40387
    },
    {
      "epoch": 16.135836995605274,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 40388
    },
    {
      "epoch": 16.136236516180585,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 40389
    },
    {
      "epoch": 16.136636036755892,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 40390
    },
    {
      "epoch": 16.137035557331203,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 40391
    },
    {
      "epoch": 16.137435077906513,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 40392
    },
    {
      "epoch": 16.13783459848182,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 40393
    },
    {
      "epoch": 16.13823411905713,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 40394
    },
    {
      "epoch": 16.13863363963244,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 40395
    },
    {
      "epoch": 16.139033160207752,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 40396
    },
    {
      "epoch": 16.13943268078306,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 40397
    },
    {
      "epoch": 16.13983220135837,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 40398
    },
    {
      "epoch": 16.14023172193368,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 40399
    },
    {
      "epoch": 16.140631242508988,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 40400
    },
    {
      "epoch": 16.1410307630843,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 40401
    },
    {
      "epoch": 16.14143028365961,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 40402
    },
    {
      "epoch": 16.141829804234916,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0135,
      "step": 40403
    },
    {
      "epoch": 16.142229324810227,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 40404
    },
    {
      "epoch": 16.142628845385538,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 40405
    },
    {
      "epoch": 16.14302836596085,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 40406
    },
    {
      "epoch": 16.143427886536156,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 40407
    },
    {
      "epoch": 16.143827407111466,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 40408
    },
    {
      "epoch": 16.144226927686777,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 40409
    },
    {
      "epoch": 16.144626448262084,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 40410
    },
    {
      "epoch": 16.145025968837395,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 40411
    },
    {
      "epoch": 16.145425489412705,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 40412
    },
    {
      "epoch": 16.145825009988016,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 40413
    },
    {
      "epoch": 16.146224530563323,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 40414
    },
    {
      "epoch": 16.146624051138634,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 40415
    },
    {
      "epoch": 16.147023571713945,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 40416
    },
    {
      "epoch": 16.14742309228925,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 40417
    },
    {
      "epoch": 16.147822612864562,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 40418
    },
    {
      "epoch": 16.148222133439873,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 40419
    },
    {
      "epoch": 16.14862165401518,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 40420
    },
    {
      "epoch": 16.14902117459049,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 40421
    },
    {
      "epoch": 16.1494206951658,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 40422
    },
    {
      "epoch": 16.149820215741112,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 40423
    },
    {
      "epoch": 16.15021973631642,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 40424
    },
    {
      "epoch": 16.15061925689173,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 40425
    },
    {
      "epoch": 16.15101877746704,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 40426
    },
    {
      "epoch": 16.151418298042348,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 40427
    },
    {
      "epoch": 16.15181781861766,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 40428
    },
    {
      "epoch": 16.15221733919297,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 40429
    },
    {
      "epoch": 16.15261685976828,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 40430
    },
    {
      "epoch": 16.153016380343587,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 40431
    },
    {
      "epoch": 16.153415900918898,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 40432
    },
    {
      "epoch": 16.15381542149421,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 40433
    },
    {
      "epoch": 16.154214942069515,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 40434
    },
    {
      "epoch": 16.154614462644826,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 40435
    },
    {
      "epoch": 16.155013983220137,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 40436
    },
    {
      "epoch": 16.155413503795444,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 40437
    },
    {
      "epoch": 16.155813024370755,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 40438
    },
    {
      "epoch": 16.156212544946065,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 40439
    },
    {
      "epoch": 16.156612065521376,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 40440
    },
    {
      "epoch": 16.157011586096683,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 40441
    },
    {
      "epoch": 16.157411106671994,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 40442
    },
    {
      "epoch": 16.157810627247304,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 40443
    },
    {
      "epoch": 16.15821014782261,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 40444
    },
    {
      "epoch": 16.158609668397922,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0077,
      "step": 40445
    },
    {
      "epoch": 16.159009188973233,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 40446
    },
    {
      "epoch": 16.159408709548543,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 40447
    },
    {
      "epoch": 16.15980823012385,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 40448
    },
    {
      "epoch": 16.16020775069916,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 40449
    },
    {
      "epoch": 16.160607271274472,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 40450
    },
    {
      "epoch": 16.16100679184978,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 40451
    },
    {
      "epoch": 16.16140631242509,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 40452
    },
    {
      "epoch": 16.1618058330004,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 40453
    },
    {
      "epoch": 16.162205353575708,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 40454
    },
    {
      "epoch": 16.162604874151018,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 40455
    },
    {
      "epoch": 16.16300439472633,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 40456
    },
    {
      "epoch": 16.16340391530164,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 40457
    },
    {
      "epoch": 16.163803435876947,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 40458
    },
    {
      "epoch": 16.164202956452257,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 40459
    },
    {
      "epoch": 16.164602477027568,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 40460
    },
    {
      "epoch": 16.165001997602875,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 40461
    },
    {
      "epoch": 16.165401518178186,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 40462
    },
    {
      "epoch": 16.165801038753496,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 40463
    },
    {
      "epoch": 16.166200559328807,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 40464
    },
    {
      "epoch": 16.166600079904114,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 40465
    },
    {
      "epoch": 16.166999600479425,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 40466
    },
    {
      "epoch": 16.167399121054736,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 40467
    },
    {
      "epoch": 16.167798641630043,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 40468
    },
    {
      "epoch": 16.168198162205353,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 40469
    },
    {
      "epoch": 16.168597682780664,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 40470
    },
    {
      "epoch": 16.16899720335597,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 40471
    },
    {
      "epoch": 16.169396723931282,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 40472
    },
    {
      "epoch": 16.169796244506593,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 40473
    },
    {
      "epoch": 16.170195765081903,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 40474
    },
    {
      "epoch": 16.17059528565721,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 40475
    },
    {
      "epoch": 16.17099480623252,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 40476
    },
    {
      "epoch": 16.17139432680783,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 40477
    },
    {
      "epoch": 16.17179384738314,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 40478
    },
    {
      "epoch": 16.17219336795845,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 40479
    },
    {
      "epoch": 16.17259288853376,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 40480
    },
    {
      "epoch": 16.17299240910907,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 40481
    },
    {
      "epoch": 16.173391929684378,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 40482
    },
    {
      "epoch": 16.17379145025969,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 40483
    },
    {
      "epoch": 16.174190970835,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 40484
    },
    {
      "epoch": 16.174590491410306,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 40485
    },
    {
      "epoch": 16.174990011985617,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 40486
    },
    {
      "epoch": 16.175389532560928,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 40487
    },
    {
      "epoch": 16.175789053136235,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 40488
    },
    {
      "epoch": 16.176188573711546,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 40489
    },
    {
      "epoch": 16.176588094286856,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 40490
    },
    {
      "epoch": 16.176987614862167,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 40491
    },
    {
      "epoch": 16.177387135437474,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 40492
    },
    {
      "epoch": 16.177786656012785,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 40493
    },
    {
      "epoch": 16.178186176588095,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 40494
    },
    {
      "epoch": 16.178585697163403,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 40495
    },
    {
      "epoch": 16.178985217738713,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 40496
    },
    {
      "epoch": 16.179384738314024,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 40497
    },
    {
      "epoch": 16.179784258889335,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 40498
    },
    {
      "epoch": 16.18018377946464,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 40499
    },
    {
      "epoch": 16.180583300039952,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 40500
    },
    {
      "epoch": 16.180982820615263,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 40501
    },
    {
      "epoch": 16.18138234119057,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 40502
    },
    {
      "epoch": 16.18178186176588,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 40503
    },
    {
      "epoch": 16.18218138234119,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 40504
    },
    {
      "epoch": 16.1825809029165,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 40505
    },
    {
      "epoch": 16.18298042349181,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 40506
    },
    {
      "epoch": 16.18337994406712,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 40507
    },
    {
      "epoch": 16.18377946464243,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 40508
    },
    {
      "epoch": 16.184178985217738,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 40509
    },
    {
      "epoch": 16.18457850579305,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 40510
    },
    {
      "epoch": 16.18497802636836,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 40511
    },
    {
      "epoch": 16.185377546943666,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 40512
    },
    {
      "epoch": 16.185777067518977,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 40513
    },
    {
      "epoch": 16.186176588094288,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 40514
    },
    {
      "epoch": 16.186576108669595,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 40515
    },
    {
      "epoch": 16.186975629244905,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 40516
    },
    {
      "epoch": 16.187375149820216,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 40517
    },
    {
      "epoch": 16.187774670395527,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 40518
    },
    {
      "epoch": 16.188174190970834,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 40519
    },
    {
      "epoch": 16.188573711546145,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 40520
    },
    {
      "epoch": 16.188973232121455,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 40521
    },
    {
      "epoch": 16.189372752696762,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 40522
    },
    {
      "epoch": 16.189772273272073,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 40523
    },
    {
      "epoch": 16.190171793847384,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 40524
    },
    {
      "epoch": 16.190571314422694,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 40525
    },
    {
      "epoch": 16.190970834998,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 40526
    },
    {
      "epoch": 16.191370355573312,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 40527
    },
    {
      "epoch": 16.191769876148623,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 40528
    },
    {
      "epoch": 16.19216939672393,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 40529
    },
    {
      "epoch": 16.19256891729924,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 40530
    },
    {
      "epoch": 16.19296843787455,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 40531
    },
    {
      "epoch": 16.19336795844986,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 40532
    },
    {
      "epoch": 16.19376747902517,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 40533
    },
    {
      "epoch": 16.19416699960048,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 40534
    },
    {
      "epoch": 16.19456652017579,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 40535
    },
    {
      "epoch": 16.194966040751098,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 40536
    },
    {
      "epoch": 16.19536556132641,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 40537
    },
    {
      "epoch": 16.19576508190172,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 40538
    },
    {
      "epoch": 16.196164602477026,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 40539
    },
    {
      "epoch": 16.196564123052337,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0099,
      "step": 40540
    },
    {
      "epoch": 16.196963643627647,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 40541
    },
    {
      "epoch": 16.197363164202958,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 40542
    },
    {
      "epoch": 16.197762684778265,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 40543
    },
    {
      "epoch": 16.198162205353576,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 40544
    },
    {
      "epoch": 16.198561725928887,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 40545
    },
    {
      "epoch": 16.198961246504194,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 40546
    },
    {
      "epoch": 16.199360767079504,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 40547
    },
    {
      "epoch": 16.199760287654815,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 40548
    },
    {
      "epoch": 16.200159808230122,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 40549
    },
    {
      "epoch": 16.200559328805433,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 40550
    },
    {
      "epoch": 16.200958849380743,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 40551
    },
    {
      "epoch": 16.201358369956054,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 40552
    },
    {
      "epoch": 16.20175789053136,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 40553
    },
    {
      "epoch": 16.202157411106672,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 40554
    },
    {
      "epoch": 16.202556931681983,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 40555
    },
    {
      "epoch": 16.20295645225729,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 40556
    },
    {
      "epoch": 16.2033559728326,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 40557
    },
    {
      "epoch": 16.20375549340791,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 40558
    },
    {
      "epoch": 16.204155013983222,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 40559
    },
    {
      "epoch": 16.20455453455853,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0078,
      "step": 40560
    },
    {
      "epoch": 16.20495405513384,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 40561
    },
    {
      "epoch": 16.20535357570915,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 40562
    },
    {
      "epoch": 16.205753096284457,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 40563
    },
    {
      "epoch": 16.206152616859768,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 40564
    },
    {
      "epoch": 16.20655213743508,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 40565
    },
    {
      "epoch": 16.206951658010386,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 40566
    },
    {
      "epoch": 16.207351178585697,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 40567
    },
    {
      "epoch": 16.207750699161007,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 40568
    },
    {
      "epoch": 16.208150219736318,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 40569
    },
    {
      "epoch": 16.208549740311625,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 40570
    },
    {
      "epoch": 16.208949260886936,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 40571
    },
    {
      "epoch": 16.209348781462246,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 40572
    },
    {
      "epoch": 16.209748302037553,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 40573
    },
    {
      "epoch": 16.210147822612864,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 40574
    },
    {
      "epoch": 16.210547343188175,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 40575
    },
    {
      "epoch": 16.210946863763485,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 40576
    },
    {
      "epoch": 16.211346384338793,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 40577
    },
    {
      "epoch": 16.211745904914103,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 40578
    },
    {
      "epoch": 16.212145425489414,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 40579
    },
    {
      "epoch": 16.21254494606472,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 40580
    },
    {
      "epoch": 16.21294446664003,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 40581
    },
    {
      "epoch": 16.213343987215342,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 40582
    },
    {
      "epoch": 16.21374350779065,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 40583
    },
    {
      "epoch": 16.21414302836596,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 40584
    },
    {
      "epoch": 16.21454254894127,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 40585
    },
    {
      "epoch": 16.21494206951658,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 40586
    },
    {
      "epoch": 16.21534159009189,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 40587
    },
    {
      "epoch": 16.2157411106672,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 40588
    },
    {
      "epoch": 16.21614063124251,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 40589
    },
    {
      "epoch": 16.216540151817817,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 40590
    },
    {
      "epoch": 16.216939672393128,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 40591
    },
    {
      "epoch": 16.21733919296844,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 40592
    },
    {
      "epoch": 16.21773871354375,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 40593
    },
    {
      "epoch": 16.218138234119056,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 40594
    },
    {
      "epoch": 16.218537754694367,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 40595
    },
    {
      "epoch": 16.218937275269678,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 40596
    },
    {
      "epoch": 16.219336795844985,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 40597
    },
    {
      "epoch": 16.219736316420295,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 40598
    },
    {
      "epoch": 16.220135836995606,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 40599
    },
    {
      "epoch": 16.220535357570913,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 40600
    },
    {
      "epoch": 16.220934878146224,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 40601
    },
    {
      "epoch": 16.221334398721535,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 40602
    },
    {
      "epoch": 16.221733919296845,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0111,
      "step": 40603
    },
    {
      "epoch": 16.222133439872152,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 40604
    },
    {
      "epoch": 16.222532960447463,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 40605
    },
    {
      "epoch": 16.222932481022774,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 40606
    },
    {
      "epoch": 16.22333200159808,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 40607
    },
    {
      "epoch": 16.22373152217339,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 40608
    },
    {
      "epoch": 16.224131042748702,
      "grad_norm": 0.036376953125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 40609
    },
    {
      "epoch": 16.224530563324013,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 40610
    },
    {
      "epoch": 16.22493008389932,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 40611
    },
    {
      "epoch": 16.22532960447463,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 40612
    },
    {
      "epoch": 16.22572912504994,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 40613
    },
    {
      "epoch": 16.22612864562525,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 40614
    },
    {
      "epoch": 16.22652816620056,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 40615
    },
    {
      "epoch": 16.22692768677587,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 40616
    },
    {
      "epoch": 16.227327207351177,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 40617
    },
    {
      "epoch": 16.227726727926488,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 40618
    },
    {
      "epoch": 16.2281262485018,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 40619
    },
    {
      "epoch": 16.22852576907711,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 40620
    },
    {
      "epoch": 16.228925289652416,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 40621
    },
    {
      "epoch": 16.229324810227727,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 40622
    },
    {
      "epoch": 16.229724330803037,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 40623
    },
    {
      "epoch": 16.230123851378345,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 40624
    },
    {
      "epoch": 16.230523371953655,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 40625
    },
    {
      "epoch": 16.230922892528966,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 40626
    },
    {
      "epoch": 16.231322413104277,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 40627
    },
    {
      "epoch": 16.231721933679584,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 40628
    },
    {
      "epoch": 16.232121454254894,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 40629
    },
    {
      "epoch": 16.232520974830205,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 40630
    },
    {
      "epoch": 16.232920495405512,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 40631
    },
    {
      "epoch": 16.233320015980823,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 40632
    },
    {
      "epoch": 16.233719536556134,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 40633
    },
    {
      "epoch": 16.23411905713144,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 40634
    },
    {
      "epoch": 16.23451857770675,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 40635
    },
    {
      "epoch": 16.234918098282062,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 40636
    },
    {
      "epoch": 16.235317618857373,
      "grad_norm": 0.03564453125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 40637
    },
    {
      "epoch": 16.23571713943268,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 40638
    },
    {
      "epoch": 16.23611666000799,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 40639
    },
    {
      "epoch": 16.2365161805833,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 40640
    },
    {
      "epoch": 16.23691570115861,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 40641
    },
    {
      "epoch": 16.23731522173392,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 40642
    },
    {
      "epoch": 16.23771474230923,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 40643
    },
    {
      "epoch": 16.23811426288454,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 40644
    },
    {
      "epoch": 16.238513783459847,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 40645
    },
    {
      "epoch": 16.238913304035158,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 40646
    },
    {
      "epoch": 16.23931282461047,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 40647
    },
    {
      "epoch": 16.239712345185776,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 40648
    },
    {
      "epoch": 16.240111865761087,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 40649
    },
    {
      "epoch": 16.240511386336397,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 40650
    },
    {
      "epoch": 16.240910906911704,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 40651
    },
    {
      "epoch": 16.241310427487015,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 40652
    },
    {
      "epoch": 16.241709948062326,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 40653
    },
    {
      "epoch": 16.242109468637636,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 40654
    },
    {
      "epoch": 16.242508989212944,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 40655
    },
    {
      "epoch": 16.242908509788254,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 40656
    },
    {
      "epoch": 16.243308030363565,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 40657
    },
    {
      "epoch": 16.243707550938872,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 40658
    },
    {
      "epoch": 16.244107071514183,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 40659
    },
    {
      "epoch": 16.244506592089493,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 40660
    },
    {
      "epoch": 16.244906112664804,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 40661
    },
    {
      "epoch": 16.24530563324011,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 40662
    },
    {
      "epoch": 16.245705153815422,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 40663
    },
    {
      "epoch": 16.246104674390732,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 40664
    },
    {
      "epoch": 16.24650419496604,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 40665
    },
    {
      "epoch": 16.24690371554135,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 40666
    },
    {
      "epoch": 16.24730323611666,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 40667
    },
    {
      "epoch": 16.247702756691968,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 40668
    },
    {
      "epoch": 16.24810227726728,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 40669
    },
    {
      "epoch": 16.24850179784259,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 40670
    },
    {
      "epoch": 16.2489013184179,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 40671
    },
    {
      "epoch": 16.249300838993207,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 40672
    },
    {
      "epoch": 16.249700359568518,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 40673
    },
    {
      "epoch": 16.25009988014383,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 40674
    },
    {
      "epoch": 16.250499400719136,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 40675
    },
    {
      "epoch": 16.250898921294446,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 40676
    },
    {
      "epoch": 16.251298441869757,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 40677
    },
    {
      "epoch": 16.251697962445064,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 40678
    },
    {
      "epoch": 16.252097483020375,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 40679
    },
    {
      "epoch": 16.252497003595685,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 40680
    },
    {
      "epoch": 16.252896524170996,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 40681
    },
    {
      "epoch": 16.253296044746303,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 40682
    },
    {
      "epoch": 16.253695565321614,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 40683
    },
    {
      "epoch": 16.254095085896925,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 40684
    },
    {
      "epoch": 16.25449460647223,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 40685
    },
    {
      "epoch": 16.254894127047542,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 40686
    },
    {
      "epoch": 16.255293647622853,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 40687
    },
    {
      "epoch": 16.255693168198164,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 40688
    },
    {
      "epoch": 16.25609268877347,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 40689
    },
    {
      "epoch": 16.25649220934878,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 40690
    },
    {
      "epoch": 16.256891729924092,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 40691
    },
    {
      "epoch": 16.2572912504994,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 40692
    },
    {
      "epoch": 16.25769077107471,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 40693
    },
    {
      "epoch": 16.25809029165002,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 40694
    },
    {
      "epoch": 16.258489812225328,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 40695
    },
    {
      "epoch": 16.25888933280064,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 40696
    },
    {
      "epoch": 16.25928885337595,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 40697
    },
    {
      "epoch": 16.25968837395126,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 40698
    },
    {
      "epoch": 16.260087894526567,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 40699
    },
    {
      "epoch": 16.260487415101878,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 40700
    },
    {
      "epoch": 16.26088693567719,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 40701
    },
    {
      "epoch": 16.261286456252495,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 40702
    },
    {
      "epoch": 16.261685976827806,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 40703
    },
    {
      "epoch": 16.262085497403117,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 40704
    },
    {
      "epoch": 16.262485017978427,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 40705
    },
    {
      "epoch": 16.262884538553735,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 40706
    },
    {
      "epoch": 16.263284059129045,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 40707
    },
    {
      "epoch": 16.263683579704356,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 40708
    },
    {
      "epoch": 16.264083100279663,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 40709
    },
    {
      "epoch": 16.264482620854974,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 40710
    },
    {
      "epoch": 16.264882141430284,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 40711
    },
    {
      "epoch": 16.26528166200559,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 40712
    },
    {
      "epoch": 16.265681182580902,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 40713
    },
    {
      "epoch": 16.266080703156213,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 40714
    },
    {
      "epoch": 16.266480223731524,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 40715
    },
    {
      "epoch": 16.26687974430683,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 40716
    },
    {
      "epoch": 16.26727926488214,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 40717
    },
    {
      "epoch": 16.267678785457452,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 40718
    },
    {
      "epoch": 16.26807830603276,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 40719
    },
    {
      "epoch": 16.26847782660807,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 40720
    },
    {
      "epoch": 16.26887734718338,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 40721
    },
    {
      "epoch": 16.26927686775869,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 40722
    },
    {
      "epoch": 16.269676388334,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 40723
    },
    {
      "epoch": 16.27007590890931,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 40724
    },
    {
      "epoch": 16.27047542948462,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 40725
    },
    {
      "epoch": 16.270874950059927,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 40726
    },
    {
      "epoch": 16.271274470635237,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 40727
    },
    {
      "epoch": 16.271673991210548,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 40728
    },
    {
      "epoch": 16.272073511785855,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 40729
    },
    {
      "epoch": 16.272473032361166,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 40730
    },
    {
      "epoch": 16.272872552936477,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 40731
    },
    {
      "epoch": 16.273272073511787,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 40732
    },
    {
      "epoch": 16.273671594087094,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0112,
      "step": 40733
    },
    {
      "epoch": 16.274071114662405,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 40734
    },
    {
      "epoch": 16.274470635237716,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 40735
    },
    {
      "epoch": 16.274870155813023,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 40736
    },
    {
      "epoch": 16.275269676388334,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 40737
    },
    {
      "epoch": 16.275669196963644,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 40738
    },
    {
      "epoch": 16.276068717538955,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 40739
    },
    {
      "epoch": 16.276468238114262,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 40740
    },
    {
      "epoch": 16.276867758689573,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 40741
    },
    {
      "epoch": 16.277267279264883,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 40742
    },
    {
      "epoch": 16.27766679984019,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 40743
    },
    {
      "epoch": 16.2780663204155,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 40744
    },
    {
      "epoch": 16.278465840990812,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 40745
    },
    {
      "epoch": 16.27886536156612,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 40746
    },
    {
      "epoch": 16.27926488214143,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 40747
    },
    {
      "epoch": 16.27966440271674,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 40748
    },
    {
      "epoch": 16.28006392329205,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 40749
    },
    {
      "epoch": 16.280463443867358,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 40750
    },
    {
      "epoch": 16.28086296444267,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 40751
    },
    {
      "epoch": 16.28126248501798,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 40752
    },
    {
      "epoch": 16.281662005593287,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 40753
    },
    {
      "epoch": 16.282061526168597,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 40754
    },
    {
      "epoch": 16.282461046743908,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 40755
    },
    {
      "epoch": 16.28286056731922,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 40756
    },
    {
      "epoch": 16.283260087894526,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 40757
    },
    {
      "epoch": 16.283659608469836,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 40758
    },
    {
      "epoch": 16.284059129045147,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 40759
    },
    {
      "epoch": 16.284458649620454,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 40760
    },
    {
      "epoch": 16.284858170195765,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 40761
    },
    {
      "epoch": 16.285257690771076,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 40762
    },
    {
      "epoch": 16.285657211346383,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 40763
    },
    {
      "epoch": 16.286056731921693,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 40764
    },
    {
      "epoch": 16.286456252497004,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 40765
    },
    {
      "epoch": 16.286855773072315,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 40766
    },
    {
      "epoch": 16.287255293647622,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 40767
    },
    {
      "epoch": 16.287654814222932,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 40768
    },
    {
      "epoch": 16.288054334798243,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 40769
    },
    {
      "epoch": 16.28845385537355,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 40770
    },
    {
      "epoch": 16.28885337594886,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 40771
    },
    {
      "epoch": 16.28925289652417,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 40772
    },
    {
      "epoch": 16.289652417099482,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 40773
    },
    {
      "epoch": 16.29005193767479,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 40774
    },
    {
      "epoch": 16.2904514582501,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 40775
    },
    {
      "epoch": 16.29085097882541,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 40776
    },
    {
      "epoch": 16.291250499400718,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 40777
    },
    {
      "epoch": 16.29165001997603,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 40778
    },
    {
      "epoch": 16.29204954055134,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 40779
    },
    {
      "epoch": 16.292449061126646,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 40780
    },
    {
      "epoch": 16.292848581701957,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 40781
    },
    {
      "epoch": 16.293248102277268,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 40782
    },
    {
      "epoch": 16.29364762285258,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 40783
    },
    {
      "epoch": 16.294047143427886,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 40784
    },
    {
      "epoch": 16.294446664003196,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 40785
    },
    {
      "epoch": 16.294846184578507,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 40786
    },
    {
      "epoch": 16.295245705153814,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 40787
    },
    {
      "epoch": 16.295645225729125,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 40788
    },
    {
      "epoch": 16.296044746304435,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 40789
    },
    {
      "epoch": 16.296444266879746,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 40790
    },
    {
      "epoch": 16.296843787455053,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 40791
    },
    {
      "epoch": 16.297243308030364,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 40792
    },
    {
      "epoch": 16.297642828605674,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 40793
    },
    {
      "epoch": 16.29804234918098,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 40794
    },
    {
      "epoch": 16.298441869756292,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 40795
    },
    {
      "epoch": 16.298841390331603,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 40796
    },
    {
      "epoch": 16.29924091090691,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 40797
    },
    {
      "epoch": 16.29964043148222,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 40798
    },
    {
      "epoch": 16.30003995205753,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 40799
    },
    {
      "epoch": 16.300439472632842,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 40800
    },
    {
      "epoch": 16.30083899320815,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0121,
      "step": 40801
    },
    {
      "epoch": 16.30123851378346,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 40802
    },
    {
      "epoch": 16.30163803435877,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 40803
    },
    {
      "epoch": 16.302037554934078,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 40804
    },
    {
      "epoch": 16.30243707550939,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 40805
    },
    {
      "epoch": 16.3028365960847,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 40806
    },
    {
      "epoch": 16.30323611666001,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 40807
    },
    {
      "epoch": 16.303635637235317,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 40808
    },
    {
      "epoch": 16.304035157810628,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 40809
    },
    {
      "epoch": 16.304434678385938,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 40810
    },
    {
      "epoch": 16.304834198961245,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 40811
    },
    {
      "epoch": 16.305233719536556,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 40812
    },
    {
      "epoch": 16.305633240111867,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 40813
    },
    {
      "epoch": 16.306032760687174,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 40814
    },
    {
      "epoch": 16.306432281262484,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 40815
    },
    {
      "epoch": 16.306831801837795,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 40816
    },
    {
      "epoch": 16.307231322413106,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 40817
    },
    {
      "epoch": 16.307630842988413,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 40818
    },
    {
      "epoch": 16.308030363563724,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 40819
    },
    {
      "epoch": 16.308429884139034,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 40820
    },
    {
      "epoch": 16.30882940471434,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 40821
    },
    {
      "epoch": 16.309228925289652,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 40822
    },
    {
      "epoch": 16.309628445864963,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 40823
    },
    {
      "epoch": 16.310027966440273,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 40824
    },
    {
      "epoch": 16.31042748701558,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 40825
    },
    {
      "epoch": 16.31082700759089,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 40826
    },
    {
      "epoch": 16.311226528166202,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 40827
    },
    {
      "epoch": 16.31162604874151,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 40828
    },
    {
      "epoch": 16.31202556931682,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 40829
    },
    {
      "epoch": 16.31242508989213,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 40830
    },
    {
      "epoch": 16.312824610467437,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 40831
    },
    {
      "epoch": 16.313224131042748,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 40832
    },
    {
      "epoch": 16.31362365161806,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 40833
    },
    {
      "epoch": 16.31402317219337,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 40834
    },
    {
      "epoch": 16.314422692768677,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 40835
    },
    {
      "epoch": 16.314822213343987,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 40836
    },
    {
      "epoch": 16.315221733919298,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 40837
    },
    {
      "epoch": 16.315621254494605,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 40838
    },
    {
      "epoch": 16.316020775069916,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 40839
    },
    {
      "epoch": 16.316420295645226,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 40840
    },
    {
      "epoch": 16.316819816220537,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 40841
    },
    {
      "epoch": 16.317219336795844,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 40842
    },
    {
      "epoch": 16.317618857371155,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0106,
      "step": 40843
    },
    {
      "epoch": 16.318018377946466,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 40844
    },
    {
      "epoch": 16.318417898521773,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 40845
    },
    {
      "epoch": 16.318817419097083,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 40846
    },
    {
      "epoch": 16.319216939672394,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 40847
    },
    {
      "epoch": 16.3196164602477,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 40848
    },
    {
      "epoch": 16.320015980823012,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 40849
    },
    {
      "epoch": 16.320415501398323,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 40850
    },
    {
      "epoch": 16.320815021973633,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 40851
    },
    {
      "epoch": 16.32121454254894,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 40852
    },
    {
      "epoch": 16.32161406312425,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 40853
    },
    {
      "epoch": 16.32201358369956,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 40854
    },
    {
      "epoch": 16.32241310427487,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 40855
    },
    {
      "epoch": 16.32281262485018,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 40856
    },
    {
      "epoch": 16.32321214542549,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 40857
    },
    {
      "epoch": 16.3236116660008,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 40858
    },
    {
      "epoch": 16.324011186576108,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 40859
    },
    {
      "epoch": 16.32441070715142,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 40860
    },
    {
      "epoch": 16.32481022772673,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 40861
    },
    {
      "epoch": 16.325209748302036,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 40862
    },
    {
      "epoch": 16.325609268877347,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 40863
    },
    {
      "epoch": 16.326008789452658,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 40864
    },
    {
      "epoch": 16.326408310027965,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 40865
    },
    {
      "epoch": 16.326807830603276,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 40866
    },
    {
      "epoch": 16.327207351178586,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 40867
    },
    {
      "epoch": 16.327606871753897,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 40868
    },
    {
      "epoch": 16.328006392329204,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 40869
    },
    {
      "epoch": 16.328405912904515,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 40870
    },
    {
      "epoch": 16.328805433479825,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0101,
      "step": 40871
    },
    {
      "epoch": 16.329204954055133,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 40872
    },
    {
      "epoch": 16.329604474630443,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 40873
    },
    {
      "epoch": 16.330003995205754,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 40874
    },
    {
      "epoch": 16.330403515781065,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 40875
    },
    {
      "epoch": 16.33080303635637,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 40876
    },
    {
      "epoch": 16.331202556931682,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 40877
    },
    {
      "epoch": 16.331602077506993,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.01,
      "step": 40878
    },
    {
      "epoch": 16.3320015980823,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 40879
    },
    {
      "epoch": 16.33240111865761,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 40880
    },
    {
      "epoch": 16.33280063923292,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 40881
    },
    {
      "epoch": 16.33320015980823,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 40882
    },
    {
      "epoch": 16.33359968038354,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 40883
    },
    {
      "epoch": 16.33399920095885,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 40884
    },
    {
      "epoch": 16.33439872153416,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 40885
    },
    {
      "epoch": 16.334798242109468,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 40886
    },
    {
      "epoch": 16.33519776268478,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 40887
    },
    {
      "epoch": 16.33559728326009,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 40888
    },
    {
      "epoch": 16.335996803835396,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 40889
    },
    {
      "epoch": 16.336396324410707,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 40890
    },
    {
      "epoch": 16.336795844986018,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 40891
    },
    {
      "epoch": 16.337195365561325,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 40892
    },
    {
      "epoch": 16.337594886136635,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 40893
    },
    {
      "epoch": 16.337994406711946,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 40894
    },
    {
      "epoch": 16.338393927287257,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 40895
    },
    {
      "epoch": 16.338793447862564,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 40896
    },
    {
      "epoch": 16.339192968437874,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 40897
    },
    {
      "epoch": 16.339592489013185,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 40898
    },
    {
      "epoch": 16.339992009588492,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 40899
    },
    {
      "epoch": 16.340391530163803,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 40900
    },
    {
      "epoch": 16.340791050739114,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 40901
    },
    {
      "epoch": 16.341190571314424,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 40902
    },
    {
      "epoch": 16.34159009188973,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 40903
    },
    {
      "epoch": 16.341989612465042,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 40904
    },
    {
      "epoch": 16.342389133040353,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 40905
    },
    {
      "epoch": 16.34278865361566,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 40906
    },
    {
      "epoch": 16.34318817419097,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 40907
    },
    {
      "epoch": 16.34358769476628,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0055,
      "step": 40908
    },
    {
      "epoch": 16.34398721534159,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 40909
    },
    {
      "epoch": 16.3443867359169,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 40910
    },
    {
      "epoch": 16.34478625649221,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 40911
    },
    {
      "epoch": 16.34518577706752,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 40912
    },
    {
      "epoch": 16.345585297642828,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 40913
    },
    {
      "epoch": 16.345984818218138,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 40914
    },
    {
      "epoch": 16.34638433879345,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0115,
      "step": 40915
    },
    {
      "epoch": 16.346783859368756,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 40916
    },
    {
      "epoch": 16.347183379944067,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 40917
    },
    {
      "epoch": 16.347582900519377,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 40918
    },
    {
      "epoch": 16.347982421094688,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 40919
    },
    {
      "epoch": 16.348381941669995,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 40920
    },
    {
      "epoch": 16.348781462245306,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 40921
    },
    {
      "epoch": 16.349180982820616,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 40922
    },
    {
      "epoch": 16.349580503395924,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 40923
    },
    {
      "epoch": 16.349980023971234,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 40924
    },
    {
      "epoch": 16.350379544546545,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 40925
    },
    {
      "epoch": 16.350779065121852,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 40926
    },
    {
      "epoch": 16.351178585697163,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 40927
    },
    {
      "epoch": 16.351578106272473,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 40928
    },
    {
      "epoch": 16.351977626847784,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 40929
    },
    {
      "epoch": 16.35237714742309,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 40930
    },
    {
      "epoch": 16.352776667998402,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 40931
    },
    {
      "epoch": 16.353176188573713,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 40932
    },
    {
      "epoch": 16.35357570914902,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 40933
    },
    {
      "epoch": 16.35397522972433,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 40934
    },
    {
      "epoch": 16.35437475029964,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 40935
    },
    {
      "epoch": 16.35477427087495,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 40936
    },
    {
      "epoch": 16.35517379145026,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 40937
    },
    {
      "epoch": 16.35557331202557,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 40938
    },
    {
      "epoch": 16.35597283260088,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 40939
    },
    {
      "epoch": 16.356372353176187,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 40940
    },
    {
      "epoch": 16.356771873751498,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 40941
    },
    {
      "epoch": 16.35717139432681,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 40942
    },
    {
      "epoch": 16.357570914902116,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 40943
    },
    {
      "epoch": 16.357970435477426,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 40944
    },
    {
      "epoch": 16.358369956052737,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 40945
    },
    {
      "epoch": 16.358769476628048,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 40946
    },
    {
      "epoch": 16.359168997203355,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 40947
    },
    {
      "epoch": 16.359568517778666,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.044,
      "step": 40948
    },
    {
      "epoch": 16.359968038353976,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 40949
    },
    {
      "epoch": 16.360367558929283,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 40950
    },
    {
      "epoch": 16.360767079504594,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 40951
    },
    {
      "epoch": 16.361166600079905,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 40952
    },
    {
      "epoch": 16.361566120655215,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 40953
    },
    {
      "epoch": 16.361965641230523,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 40954
    },
    {
      "epoch": 16.362365161805833,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 40955
    },
    {
      "epoch": 16.362764682381144,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 40956
    },
    {
      "epoch": 16.36316420295645,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 40957
    },
    {
      "epoch": 16.36356372353176,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 40958
    },
    {
      "epoch": 16.363963244107072,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 40959
    },
    {
      "epoch": 16.36436276468238,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 40960
    },
    {
      "epoch": 16.36476228525769,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 40961
    },
    {
      "epoch": 16.365161805833,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 40962
    },
    {
      "epoch": 16.36556132640831,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 40963
    },
    {
      "epoch": 16.36596084698362,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 40964
    },
    {
      "epoch": 16.36636036755893,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 40965
    },
    {
      "epoch": 16.36675988813424,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 40966
    },
    {
      "epoch": 16.367159408709547,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 40967
    },
    {
      "epoch": 16.367558929284858,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 40968
    },
    {
      "epoch": 16.36795844986017,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 40969
    },
    {
      "epoch": 16.36835797043548,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0053,
      "step": 40970
    },
    {
      "epoch": 16.368757491010786,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 40971
    },
    {
      "epoch": 16.369157011586097,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 40972
    },
    {
      "epoch": 16.369556532161408,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 40973
    },
    {
      "epoch": 16.369956052736715,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 40974
    },
    {
      "epoch": 16.370355573312025,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 40975
    },
    {
      "epoch": 16.370755093887336,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 40976
    },
    {
      "epoch": 16.371154614462643,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 40977
    },
    {
      "epoch": 16.371554135037954,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 40978
    },
    {
      "epoch": 16.371953655613265,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 40979
    },
    {
      "epoch": 16.372353176188575,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 40980
    },
    {
      "epoch": 16.372752696763882,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 40981
    },
    {
      "epoch": 16.373152217339193,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 40982
    },
    {
      "epoch": 16.373551737914504,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 40983
    },
    {
      "epoch": 16.37395125848981,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 40984
    },
    {
      "epoch": 16.37435077906512,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 40985
    },
    {
      "epoch": 16.374750299640432,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 40986
    },
    {
      "epoch": 16.375149820215743,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 40987
    },
    {
      "epoch": 16.37554934079105,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 40988
    },
    {
      "epoch": 16.37594886136636,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0055,
      "step": 40989
    },
    {
      "epoch": 16.37634838194167,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 40990
    },
    {
      "epoch": 16.37674790251698,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 40991
    },
    {
      "epoch": 16.37714742309229,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 40992
    },
    {
      "epoch": 16.3775469436676,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 40993
    },
    {
      "epoch": 16.377946464242907,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 40994
    },
    {
      "epoch": 16.378345984818218,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 40995
    },
    {
      "epoch": 16.37874550539353,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 40996
    },
    {
      "epoch": 16.37914502596884,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 40997
    },
    {
      "epoch": 16.379544546544146,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 40998
    },
    {
      "epoch": 16.379944067119457,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 40999
    },
    {
      "epoch": 16.380343587694767,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 41000
    },
    {
      "epoch": 16.380743108270075,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 41001
    },
    {
      "epoch": 16.381142628845385,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 41002
    },
    {
      "epoch": 16.381542149420696,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 41003
    },
    {
      "epoch": 16.381941669996007,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 41004
    },
    {
      "epoch": 16.382341190571314,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 41005
    },
    {
      "epoch": 16.382740711146624,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 41006
    },
    {
      "epoch": 16.383140231721935,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 41007
    },
    {
      "epoch": 16.383539752297242,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 41008
    },
    {
      "epoch": 16.383939272872553,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 41009
    },
    {
      "epoch": 16.384338793447863,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 41010
    },
    {
      "epoch": 16.38473831402317,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 41011
    },
    {
      "epoch": 16.38513783459848,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 41012
    },
    {
      "epoch": 16.385537355173792,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 41013
    },
    {
      "epoch": 16.385936875749103,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 41014
    },
    {
      "epoch": 16.38633639632441,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 41015
    },
    {
      "epoch": 16.38673591689972,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 41016
    },
    {
      "epoch": 16.38713543747503,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 41017
    },
    {
      "epoch": 16.387534958050338,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 41018
    },
    {
      "epoch": 16.38793447862565,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 41019
    },
    {
      "epoch": 16.38833399920096,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 41020
    },
    {
      "epoch": 16.388733519776267,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 41021
    },
    {
      "epoch": 16.389133040351577,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 41022
    },
    {
      "epoch": 16.389532560926888,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 41023
    },
    {
      "epoch": 16.3899320815022,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 41024
    },
    {
      "epoch": 16.390331602077506,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 41025
    },
    {
      "epoch": 16.390731122652817,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 41026
    },
    {
      "epoch": 16.391130643228127,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 41027
    },
    {
      "epoch": 16.391530163803434,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 41028
    },
    {
      "epoch": 16.391929684378745,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 41029
    },
    {
      "epoch": 16.392329204954056,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 41030
    },
    {
      "epoch": 16.392728725529366,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 41031
    },
    {
      "epoch": 16.393128246104673,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 41032
    },
    {
      "epoch": 16.393527766679984,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 41033
    },
    {
      "epoch": 16.393927287255295,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 41034
    },
    {
      "epoch": 16.394326807830602,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0111,
      "step": 41035
    },
    {
      "epoch": 16.394726328405913,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 41036
    },
    {
      "epoch": 16.395125848981223,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 41037
    },
    {
      "epoch": 16.39552536955653,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 41038
    },
    {
      "epoch": 16.39592489013184,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 41039
    },
    {
      "epoch": 16.39632441070715,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 41040
    },
    {
      "epoch": 16.396723931282462,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 41041
    },
    {
      "epoch": 16.39712345185777,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0131,
      "step": 41042
    },
    {
      "epoch": 16.39752297243308,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 41043
    },
    {
      "epoch": 16.39792249300839,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 41044
    },
    {
      "epoch": 16.398322013583698,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 41045
    },
    {
      "epoch": 16.39872153415901,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 41046
    },
    {
      "epoch": 16.39912105473432,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 41047
    },
    {
      "epoch": 16.39952057530963,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 41048
    },
    {
      "epoch": 16.399920095884937,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 41049
    },
    {
      "epoch": 16.400319616460248,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 41050
    },
    {
      "epoch": 16.40071913703556,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 41051
    },
    {
      "epoch": 16.401118657610866,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 41052
    },
    {
      "epoch": 16.401518178186176,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 41053
    },
    {
      "epoch": 16.401917698761487,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 41054
    },
    {
      "epoch": 16.402317219336794,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 41055
    },
    {
      "epoch": 16.402716739912105,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 41056
    },
    {
      "epoch": 16.403116260487415,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 41057
    },
    {
      "epoch": 16.403515781062726,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 41058
    },
    {
      "epoch": 16.403915301638033,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 41059
    },
    {
      "epoch": 16.404314822213344,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 41060
    },
    {
      "epoch": 16.404714342788655,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 41061
    },
    {
      "epoch": 16.40511386336396,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 41062
    },
    {
      "epoch": 16.405513383939272,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 41063
    },
    {
      "epoch": 16.405912904514583,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 41064
    },
    {
      "epoch": 16.406312425089894,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 41065
    },
    {
      "epoch": 16.4067119456652,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 41066
    },
    {
      "epoch": 16.40711146624051,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 41067
    },
    {
      "epoch": 16.407510986815822,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 41068
    },
    {
      "epoch": 16.40791050739113,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 41069
    },
    {
      "epoch": 16.40831002796644,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 41070
    },
    {
      "epoch": 16.40870954854175,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 41071
    },
    {
      "epoch": 16.409109069117058,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 41072
    },
    {
      "epoch": 16.40950858969237,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 41073
    },
    {
      "epoch": 16.40990811026768,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 41074
    },
    {
      "epoch": 16.41030763084299,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 41075
    },
    {
      "epoch": 16.410707151418297,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 41076
    },
    {
      "epoch": 16.411106671993608,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 41077
    },
    {
      "epoch": 16.41150619256892,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 41078
    },
    {
      "epoch": 16.411905713144225,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 41079
    },
    {
      "epoch": 16.412305233719536,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 41080
    },
    {
      "epoch": 16.412704754294847,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 41081
    },
    {
      "epoch": 16.413104274870157,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 41082
    },
    {
      "epoch": 16.413503795445465,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 41083
    },
    {
      "epoch": 16.413903316020775,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 41084
    },
    {
      "epoch": 16.414302836596086,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 41085
    },
    {
      "epoch": 16.414702357171393,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 41086
    },
    {
      "epoch": 16.415101877746704,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 41087
    },
    {
      "epoch": 16.415501398322014,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 41088
    },
    {
      "epoch": 16.41590091889732,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 41089
    },
    {
      "epoch": 16.416300439472632,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 41090
    },
    {
      "epoch": 16.416699960047943,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 41091
    },
    {
      "epoch": 16.417099480623254,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 41092
    },
    {
      "epoch": 16.41749900119856,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 41093
    },
    {
      "epoch": 16.41789852177387,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 41094
    },
    {
      "epoch": 16.418298042349182,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 41095
    },
    {
      "epoch": 16.41869756292449,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 41096
    },
    {
      "epoch": 16.4190970834998,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 41097
    },
    {
      "epoch": 16.41949660407511,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 41098
    },
    {
      "epoch": 16.41989612465042,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 41099
    },
    {
      "epoch": 16.42029564522573,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 41100
    },
    {
      "epoch": 16.42069516580104,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0135,
      "step": 41101
    },
    {
      "epoch": 16.42109468637635,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0459,
      "step": 41102
    },
    {
      "epoch": 16.421494206951657,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 41103
    },
    {
      "epoch": 16.421893727526967,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 41104
    },
    {
      "epoch": 16.422293248102278,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 41105
    },
    {
      "epoch": 16.422692768677585,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 41106
    },
    {
      "epoch": 16.423092289252896,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 41107
    },
    {
      "epoch": 16.423491809828207,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 41108
    },
    {
      "epoch": 16.423891330403517,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 41109
    },
    {
      "epoch": 16.424290850978824,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 41110
    },
    {
      "epoch": 16.424690371554135,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 41111
    },
    {
      "epoch": 16.425089892129446,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 41112
    },
    {
      "epoch": 16.425489412704753,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 41113
    },
    {
      "epoch": 16.425888933280063,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 41114
    },
    {
      "epoch": 16.426288453855374,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 41115
    },
    {
      "epoch": 16.426687974430685,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 41116
    },
    {
      "epoch": 16.427087495005992,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 41117
    },
    {
      "epoch": 16.427487015581303,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 41118
    },
    {
      "epoch": 16.427886536156613,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 41119
    },
    {
      "epoch": 16.42828605673192,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 41120
    },
    {
      "epoch": 16.42868557730723,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 41121
    },
    {
      "epoch": 16.429085097882542,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 41122
    },
    {
      "epoch": 16.42948461845785,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 41123
    },
    {
      "epoch": 16.42988413903316,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 41124
    },
    {
      "epoch": 16.43028365960847,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 41125
    },
    {
      "epoch": 16.43068318018378,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 41126
    },
    {
      "epoch": 16.431082700759088,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 41127
    },
    {
      "epoch": 16.4314822213344,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 41128
    },
    {
      "epoch": 16.43188174190971,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 41129
    },
    {
      "epoch": 16.432281262485017,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 41130
    },
    {
      "epoch": 16.432680783060327,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 41131
    },
    {
      "epoch": 16.433080303635638,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 41132
    },
    {
      "epoch": 16.43347982421095,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 41133
    },
    {
      "epoch": 16.433879344786256,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 41134
    },
    {
      "epoch": 16.434278865361566,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 41135
    },
    {
      "epoch": 16.434678385936877,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 41136
    },
    {
      "epoch": 16.435077906512184,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 41137
    },
    {
      "epoch": 16.435477427087495,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 41138
    },
    {
      "epoch": 16.435876947662805,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 41139
    },
    {
      "epoch": 16.436276468238113,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 41140
    },
    {
      "epoch": 16.436675988813423,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 41141
    },
    {
      "epoch": 16.437075509388734,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 41142
    },
    {
      "epoch": 16.437475029964045,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 41143
    },
    {
      "epoch": 16.43787455053935,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 41144
    },
    {
      "epoch": 16.438274071114662,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 41145
    },
    {
      "epoch": 16.438673591689973,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 41146
    },
    {
      "epoch": 16.43907311226528,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 41147
    },
    {
      "epoch": 16.43947263284059,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 41148
    },
    {
      "epoch": 16.4398721534159,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 41149
    },
    {
      "epoch": 16.440271673991212,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 41150
    },
    {
      "epoch": 16.44067119456652,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 41151
    },
    {
      "epoch": 16.44107071514183,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 41152
    },
    {
      "epoch": 16.44147023571714,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 41153
    },
    {
      "epoch": 16.441869756292448,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 41154
    },
    {
      "epoch": 16.44226927686776,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 41155
    },
    {
      "epoch": 16.44266879744307,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 41156
    },
    {
      "epoch": 16.443068318018376,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 41157
    },
    {
      "epoch": 16.443467838593687,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 41158
    },
    {
      "epoch": 16.443867359168998,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 41159
    },
    {
      "epoch": 16.44426687974431,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 41160
    },
    {
      "epoch": 16.444666400319615,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 41161
    },
    {
      "epoch": 16.445065920894926,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 41162
    },
    {
      "epoch": 16.445465441470237,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 41163
    },
    {
      "epoch": 16.445864962045544,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 41164
    },
    {
      "epoch": 16.446264482620855,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 41165
    },
    {
      "epoch": 16.446664003196165,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 41166
    },
    {
      "epoch": 16.447063523771476,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 41167
    },
    {
      "epoch": 16.447463044346783,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 41168
    },
    {
      "epoch": 16.447862564922094,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 41169
    },
    {
      "epoch": 16.448262085497404,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 41170
    },
    {
      "epoch": 16.44866160607271,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 41171
    },
    {
      "epoch": 16.449061126648022,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 41172
    },
    {
      "epoch": 16.449460647223333,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 41173
    },
    {
      "epoch": 16.44986016779864,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 41174
    },
    {
      "epoch": 16.45025968837395,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.008,
      "step": 41175
    },
    {
      "epoch": 16.45065920894926,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 41176
    },
    {
      "epoch": 16.451058729524572,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 41177
    },
    {
      "epoch": 16.45145825009988,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 41178
    },
    {
      "epoch": 16.45185777067519,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 41179
    },
    {
      "epoch": 16.4522572912505,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 41180
    },
    {
      "epoch": 16.452656811825808,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 41181
    },
    {
      "epoch": 16.45305633240112,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 41182
    },
    {
      "epoch": 16.45345585297643,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 41183
    },
    {
      "epoch": 16.45385537355174,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 41184
    },
    {
      "epoch": 16.454254894127047,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 41185
    },
    {
      "epoch": 16.454654414702357,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 41186
    },
    {
      "epoch": 16.455053935277668,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 41187
    },
    {
      "epoch": 16.455453455852975,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 41188
    },
    {
      "epoch": 16.455852976428286,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0078,
      "step": 41189
    },
    {
      "epoch": 16.456252497003597,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 41190
    },
    {
      "epoch": 16.456652017578904,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 41191
    },
    {
      "epoch": 16.457051538154214,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 41192
    },
    {
      "epoch": 16.457451058729525,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 41193
    },
    {
      "epoch": 16.457850579304836,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 41194
    },
    {
      "epoch": 16.458250099880143,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 41195
    },
    {
      "epoch": 16.458649620455454,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 41196
    },
    {
      "epoch": 16.459049141030764,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 41197
    },
    {
      "epoch": 16.45944866160607,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 41198
    },
    {
      "epoch": 16.459848182181382,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 41199
    },
    {
      "epoch": 16.460247702756693,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 41200
    },
    {
      "epoch": 16.460647223332003,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0079,
      "step": 41201
    },
    {
      "epoch": 16.46104674390731,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 41202
    },
    {
      "epoch": 16.46144626448262,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 41203
    },
    {
      "epoch": 16.461845785057932,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 41204
    },
    {
      "epoch": 16.46224530563324,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 41205
    },
    {
      "epoch": 16.46264482620855,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 41206
    },
    {
      "epoch": 16.46304434678386,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 41207
    },
    {
      "epoch": 16.463443867359167,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 41208
    },
    {
      "epoch": 16.463843387934478,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 41209
    },
    {
      "epoch": 16.46424290850979,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 41210
    },
    {
      "epoch": 16.4646424290851,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 41211
    },
    {
      "epoch": 16.465041949660407,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 41212
    },
    {
      "epoch": 16.465441470235717,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0421,
      "step": 41213
    },
    {
      "epoch": 16.465840990811028,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 41214
    },
    {
      "epoch": 16.466240511386335,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 41215
    },
    {
      "epoch": 16.466640031961646,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0122,
      "step": 41216
    },
    {
      "epoch": 16.467039552536956,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 41217
    },
    {
      "epoch": 16.467439073112267,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 41218
    },
    {
      "epoch": 16.467838593687574,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 41219
    },
    {
      "epoch": 16.468238114262885,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 41220
    },
    {
      "epoch": 16.468637634838196,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 41221
    },
    {
      "epoch": 16.469037155413503,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 41222
    },
    {
      "epoch": 16.469436675988813,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 41223
    },
    {
      "epoch": 16.469836196564124,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 41224
    },
    {
      "epoch": 16.47023571713943,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 41225
    },
    {
      "epoch": 16.470635237714742,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 41226
    },
    {
      "epoch": 16.471034758290052,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 41227
    },
    {
      "epoch": 16.471434278865363,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 41228
    },
    {
      "epoch": 16.47183379944067,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 41229
    },
    {
      "epoch": 16.47223332001598,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 41230
    },
    {
      "epoch": 16.47263284059129,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 41231
    },
    {
      "epoch": 16.4730323611666,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 41232
    },
    {
      "epoch": 16.47343188174191,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 41233
    },
    {
      "epoch": 16.47383140231722,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 41234
    },
    {
      "epoch": 16.474230922892527,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 41235
    },
    {
      "epoch": 16.474630443467838,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 41236
    },
    {
      "epoch": 16.47502996404315,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 41237
    },
    {
      "epoch": 16.47542948461846,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 41238
    },
    {
      "epoch": 16.475829005193766,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 41239
    },
    {
      "epoch": 16.476228525769077,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 41240
    },
    {
      "epoch": 16.476628046344388,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 41241
    },
    {
      "epoch": 16.477027566919695,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 41242
    },
    {
      "epoch": 16.477427087495006,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 41243
    },
    {
      "epoch": 16.477826608070316,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 41244
    },
    {
      "epoch": 16.478226128645627,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 41245
    },
    {
      "epoch": 16.478625649220934,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 41246
    },
    {
      "epoch": 16.479025169796245,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 41247
    },
    {
      "epoch": 16.479424690371555,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 41248
    },
    {
      "epoch": 16.479824210946862,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 41249
    },
    {
      "epoch": 16.480223731522173,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 41250
    },
    {
      "epoch": 16.480623252097484,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 41251
    },
    {
      "epoch": 16.48102277267279,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 41252
    },
    {
      "epoch": 16.4814222932481,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 41253
    },
    {
      "epoch": 16.481821813823412,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 41254
    },
    {
      "epoch": 16.482221334398723,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 41255
    },
    {
      "epoch": 16.48262085497403,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 41256
    },
    {
      "epoch": 16.48302037554934,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 41257
    },
    {
      "epoch": 16.48341989612465,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 41258
    },
    {
      "epoch": 16.48381941669996,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 41259
    },
    {
      "epoch": 16.48421893727527,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 41260
    },
    {
      "epoch": 16.48461845785058,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 41261
    },
    {
      "epoch": 16.48501797842589,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 41262
    },
    {
      "epoch": 16.485417499001198,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 41263
    },
    {
      "epoch": 16.48581701957651,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 41264
    },
    {
      "epoch": 16.48621654015182,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 41265
    },
    {
      "epoch": 16.486616060727126,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 41266
    },
    {
      "epoch": 16.487015581302437,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 41267
    },
    {
      "epoch": 16.487415101877748,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 41268
    },
    {
      "epoch": 16.487814622453055,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 41269
    },
    {
      "epoch": 16.488214143028365,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 41270
    },
    {
      "epoch": 16.488613663603676,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 41271
    },
    {
      "epoch": 16.489013184178987,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 41272
    },
    {
      "epoch": 16.489412704754294,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 41273
    },
    {
      "epoch": 16.489812225329604,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 41274
    },
    {
      "epoch": 16.490211745904915,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 41275
    },
    {
      "epoch": 16.490611266480222,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 41276
    },
    {
      "epoch": 16.491010787055533,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 41277
    },
    {
      "epoch": 16.491410307630844,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 41278
    },
    {
      "epoch": 16.491809828206154,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 41279
    },
    {
      "epoch": 16.49220934878146,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 41280
    },
    {
      "epoch": 16.492608869356772,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 41281
    },
    {
      "epoch": 16.493008389932083,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 41282
    },
    {
      "epoch": 16.49340791050739,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 41283
    },
    {
      "epoch": 16.4938074310827,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 41284
    },
    {
      "epoch": 16.49420695165801,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 41285
    },
    {
      "epoch": 16.49460647223332,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 41286
    },
    {
      "epoch": 16.49500599280863,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 41287
    },
    {
      "epoch": 16.49540551338394,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 41288
    },
    {
      "epoch": 16.49580503395925,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 41289
    },
    {
      "epoch": 16.496204554534557,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 41290
    },
    {
      "epoch": 16.496604075109868,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 41291
    },
    {
      "epoch": 16.49700359568518,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 41292
    },
    {
      "epoch": 16.497403116260486,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 41293
    },
    {
      "epoch": 16.497802636835797,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 41294
    },
    {
      "epoch": 16.498202157411107,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 41295
    },
    {
      "epoch": 16.498601677986418,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 41296
    },
    {
      "epoch": 16.499001198561725,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 41297
    },
    {
      "epoch": 16.499400719137036,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 41298
    },
    {
      "epoch": 16.499800239712346,
      "grad_norm": 0.0361328125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 41299
    },
    {
      "epoch": 16.500199760287654,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 41300
    },
    {
      "epoch": 16.500599280862964,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 41301
    },
    {
      "epoch": 16.500998801438275,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 41302
    },
    {
      "epoch": 16.501398322013582,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 41303
    },
    {
      "epoch": 16.501797842588893,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 41304
    },
    {
      "epoch": 16.502197363164203,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 41305
    },
    {
      "epoch": 16.502596883739514,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 41306
    },
    {
      "epoch": 16.50299640431482,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 41307
    },
    {
      "epoch": 16.503395924890132,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 41308
    },
    {
      "epoch": 16.503795445465443,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 41309
    },
    {
      "epoch": 16.50419496604075,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 41310
    },
    {
      "epoch": 16.50459448661606,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0057,
      "step": 41311
    },
    {
      "epoch": 16.50499400719137,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 41312
    },
    {
      "epoch": 16.50539352776668,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 41313
    },
    {
      "epoch": 16.50579304834199,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 41314
    },
    {
      "epoch": 16.5061925689173,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 41315
    },
    {
      "epoch": 16.50659208949261,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 41316
    },
    {
      "epoch": 16.506991610067917,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 41317
    },
    {
      "epoch": 16.507391130643228,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 41318
    },
    {
      "epoch": 16.50779065121854,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 41319
    },
    {
      "epoch": 16.508190171793846,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 41320
    },
    {
      "epoch": 16.508589692369156,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.012,
      "step": 41321
    },
    {
      "epoch": 16.508989212944467,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 41322
    },
    {
      "epoch": 16.509388733519778,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 41323
    },
    {
      "epoch": 16.509788254095085,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 41324
    },
    {
      "epoch": 16.510187774670396,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 41325
    },
    {
      "epoch": 16.510587295245706,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 41326
    },
    {
      "epoch": 16.510986815821013,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 41327
    },
    {
      "epoch": 16.511386336396324,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0392,
      "step": 41328
    },
    {
      "epoch": 16.511785856971635,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 41329
    },
    {
      "epoch": 16.512185377546945,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 41330
    },
    {
      "epoch": 16.512584898122252,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 41331
    },
    {
      "epoch": 16.512984418697563,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 41332
    },
    {
      "epoch": 16.513383939272874,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 41333
    },
    {
      "epoch": 16.51378345984818,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 41334
    },
    {
      "epoch": 16.51418298042349,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 41335
    },
    {
      "epoch": 16.514582500998802,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 41336
    },
    {
      "epoch": 16.51498202157411,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 41337
    },
    {
      "epoch": 16.51538154214942,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 41338
    },
    {
      "epoch": 16.51578106272473,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 41339
    },
    {
      "epoch": 16.51618058330004,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 41340
    },
    {
      "epoch": 16.51658010387535,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 41341
    },
    {
      "epoch": 16.51697962445066,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 41342
    },
    {
      "epoch": 16.51737914502597,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 41343
    },
    {
      "epoch": 16.517778665601277,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 41344
    },
    {
      "epoch": 16.518178186176588,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 41345
    },
    {
      "epoch": 16.5185777067519,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 41346
    },
    {
      "epoch": 16.518977227327206,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 41347
    },
    {
      "epoch": 16.519376747902516,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 41348
    },
    {
      "epoch": 16.519776268477827,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 41349
    },
    {
      "epoch": 16.520175789053138,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 41350
    },
    {
      "epoch": 16.520575309628445,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 41351
    },
    {
      "epoch": 16.520974830203755,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 41352
    },
    {
      "epoch": 16.521374350779066,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 41353
    },
    {
      "epoch": 16.521773871354373,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 41354
    },
    {
      "epoch": 16.522173391929684,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 41355
    },
    {
      "epoch": 16.522572912504994,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 41356
    },
    {
      "epoch": 16.522972433080305,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 41357
    },
    {
      "epoch": 16.523371953655612,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 41358
    },
    {
      "epoch": 16.523771474230923,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 41359
    },
    {
      "epoch": 16.524170994806234,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 41360
    },
    {
      "epoch": 16.52457051538154,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 41361
    },
    {
      "epoch": 16.52497003595685,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 41362
    },
    {
      "epoch": 16.525369556532162,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 41363
    },
    {
      "epoch": 16.52576907710747,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 41364
    },
    {
      "epoch": 16.52616859768278,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 41365
    },
    {
      "epoch": 16.52656811825809,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 41366
    },
    {
      "epoch": 16.5269676388334,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 41367
    },
    {
      "epoch": 16.52736715940871,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 41368
    },
    {
      "epoch": 16.52776667998402,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 41369
    },
    {
      "epoch": 16.52816620055933,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 41370
    },
    {
      "epoch": 16.528565721134637,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 41371
    },
    {
      "epoch": 16.528965241709948,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 41372
    },
    {
      "epoch": 16.529364762285258,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 41373
    },
    {
      "epoch": 16.52976428286057,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 41374
    },
    {
      "epoch": 16.530163803435876,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 41375
    },
    {
      "epoch": 16.530563324011187,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 41376
    },
    {
      "epoch": 16.530962844586497,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 41377
    },
    {
      "epoch": 16.531362365161804,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 41378
    },
    {
      "epoch": 16.531761885737115,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 41379
    },
    {
      "epoch": 16.532161406312426,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 41380
    },
    {
      "epoch": 16.532560926887733,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 41381
    },
    {
      "epoch": 16.532960447463044,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 41382
    },
    {
      "epoch": 16.533359968038354,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 41383
    },
    {
      "epoch": 16.533759488613665,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 41384
    },
    {
      "epoch": 16.534159009188972,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 41385
    },
    {
      "epoch": 16.534558529764283,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0103,
      "step": 41386
    },
    {
      "epoch": 16.534958050339593,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 41387
    },
    {
      "epoch": 16.5353575709149,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 41388
    },
    {
      "epoch": 16.53575709149021,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 41389
    },
    {
      "epoch": 16.536156612065522,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0095,
      "step": 41390
    },
    {
      "epoch": 16.536556132640833,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 41391
    },
    {
      "epoch": 16.53695565321614,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 41392
    },
    {
      "epoch": 16.53735517379145,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 41393
    },
    {
      "epoch": 16.53775469436676,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 41394
    },
    {
      "epoch": 16.538154214942068,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 41395
    },
    {
      "epoch": 16.53855373551738,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0125,
      "step": 41396
    },
    {
      "epoch": 16.53895325609269,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 41397
    },
    {
      "epoch": 16.539352776667997,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 41398
    },
    {
      "epoch": 16.539752297243307,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 41399
    },
    {
      "epoch": 16.540151817818618,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 41400
    },
    {
      "epoch": 16.54055133839393,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 41401
    },
    {
      "epoch": 16.540950858969236,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 41402
    },
    {
      "epoch": 16.541350379544546,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 41403
    },
    {
      "epoch": 16.541749900119857,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 41404
    },
    {
      "epoch": 16.542149420695164,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 41405
    },
    {
      "epoch": 16.542548941270475,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 41406
    },
    {
      "epoch": 16.542948461845786,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 41407
    },
    {
      "epoch": 16.543347982421096,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 41408
    },
    {
      "epoch": 16.543747502996403,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 41409
    },
    {
      "epoch": 16.544147023571714,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 41410
    },
    {
      "epoch": 16.544546544147025,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 41411
    },
    {
      "epoch": 16.544946064722332,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 41412
    },
    {
      "epoch": 16.545345585297643,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 41413
    },
    {
      "epoch": 16.545745105872953,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 41414
    },
    {
      "epoch": 16.54614462644826,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 41415
    },
    {
      "epoch": 16.54654414702357,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 41416
    },
    {
      "epoch": 16.54694366759888,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 41417
    },
    {
      "epoch": 16.547343188174192,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 41418
    },
    {
      "epoch": 16.5477427087495,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 41419
    },
    {
      "epoch": 16.54814222932481,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 41420
    },
    {
      "epoch": 16.54854174990012,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 41421
    },
    {
      "epoch": 16.548941270475428,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 41422
    },
    {
      "epoch": 16.54934079105074,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 41423
    },
    {
      "epoch": 16.54974031162605,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 41424
    },
    {
      "epoch": 16.55013983220136,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 41425
    },
    {
      "epoch": 16.550539352776667,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 41426
    },
    {
      "epoch": 16.550938873351978,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 41427
    },
    {
      "epoch": 16.55133839392729,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 41428
    },
    {
      "epoch": 16.551737914502596,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 41429
    },
    {
      "epoch": 16.552137435077906,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 41430
    },
    {
      "epoch": 16.552536955653217,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 41431
    },
    {
      "epoch": 16.552936476228524,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 41432
    },
    {
      "epoch": 16.553335996803835,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 41433
    },
    {
      "epoch": 16.553735517379145,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 41434
    },
    {
      "epoch": 16.554135037954456,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 41435
    },
    {
      "epoch": 16.554534558529763,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 41436
    },
    {
      "epoch": 16.554934079105074,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 41437
    },
    {
      "epoch": 16.555333599680385,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 41438
    },
    {
      "epoch": 16.55573312025569,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 41439
    },
    {
      "epoch": 16.556132640831002,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 41440
    },
    {
      "epoch": 16.556532161406313,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 41441
    },
    {
      "epoch": 16.556931681981624,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 41442
    },
    {
      "epoch": 16.55733120255693,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 41443
    },
    {
      "epoch": 16.55773072313224,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 41444
    },
    {
      "epoch": 16.558130243707552,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 41445
    },
    {
      "epoch": 16.55852976428286,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 41446
    },
    {
      "epoch": 16.55892928485817,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 41447
    },
    {
      "epoch": 16.55932880543348,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 41448
    },
    {
      "epoch": 16.559728326008788,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 41449
    },
    {
      "epoch": 16.5601278465841,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 41450
    },
    {
      "epoch": 16.56052736715941,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 41451
    },
    {
      "epoch": 16.56092688773472,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 41452
    },
    {
      "epoch": 16.561326408310027,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 41453
    },
    {
      "epoch": 16.561725928885338,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 41454
    },
    {
      "epoch": 16.56212544946065,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 41455
    },
    {
      "epoch": 16.562524970035955,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 41456
    },
    {
      "epoch": 16.562924490611266,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 41457
    },
    {
      "epoch": 16.563324011186577,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 41458
    },
    {
      "epoch": 16.563723531761887,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 41459
    },
    {
      "epoch": 16.564123052337195,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 41460
    },
    {
      "epoch": 16.564522572912505,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 41461
    },
    {
      "epoch": 16.564922093487816,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 41462
    },
    {
      "epoch": 16.565321614063123,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 41463
    },
    {
      "epoch": 16.565721134638434,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 41464
    },
    {
      "epoch": 16.566120655213744,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 41465
    },
    {
      "epoch": 16.56652017578905,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 41466
    },
    {
      "epoch": 16.566919696364362,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 41467
    },
    {
      "epoch": 16.567319216939673,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 41468
    },
    {
      "epoch": 16.567718737514983,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 41469
    },
    {
      "epoch": 16.56811825809029,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 41470
    },
    {
      "epoch": 16.5685177786656,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 41471
    },
    {
      "epoch": 16.568917299240912,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 41472
    },
    {
      "epoch": 16.56931681981622,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 41473
    },
    {
      "epoch": 16.56971634039153,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 41474
    },
    {
      "epoch": 16.57011586096684,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 41475
    },
    {
      "epoch": 16.57051538154215,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 41476
    },
    {
      "epoch": 16.570914902117458,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 41477
    },
    {
      "epoch": 16.57131442269277,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 41478
    },
    {
      "epoch": 16.57171394326808,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 41479
    },
    {
      "epoch": 16.572113463843387,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 41480
    },
    {
      "epoch": 16.572512984418697,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 41481
    },
    {
      "epoch": 16.572912504994008,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 41482
    },
    {
      "epoch": 16.573312025569315,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 41483
    },
    {
      "epoch": 16.573711546144626,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 41484
    },
    {
      "epoch": 16.574111066719937,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 41485
    },
    {
      "epoch": 16.574510587295247,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 41486
    },
    {
      "epoch": 16.574910107870554,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 41487
    },
    {
      "epoch": 16.575309628445865,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 41488
    },
    {
      "epoch": 16.575709149021176,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 41489
    },
    {
      "epoch": 16.576108669596483,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 41490
    },
    {
      "epoch": 16.576508190171793,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 41491
    },
    {
      "epoch": 16.576907710747104,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 41492
    },
    {
      "epoch": 16.577307231322415,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 41493
    },
    {
      "epoch": 16.577706751897722,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 41494
    },
    {
      "epoch": 16.578106272473033,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 41495
    },
    {
      "epoch": 16.578505793048343,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 41496
    },
    {
      "epoch": 16.57890531362365,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 41497
    },
    {
      "epoch": 16.57930483419896,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 41498
    },
    {
      "epoch": 16.57970435477427,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0059,
      "step": 41499
    },
    {
      "epoch": 16.58010387534958,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 41500
    },
    {
      "epoch": 16.58050339592489,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 41501
    },
    {
      "epoch": 16.5809029165002,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 41502
    },
    {
      "epoch": 16.58130243707551,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 41503
    },
    {
      "epoch": 16.581701957650818,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 41504
    },
    {
      "epoch": 16.58210147822613,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 41505
    },
    {
      "epoch": 16.58250099880144,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 41506
    },
    {
      "epoch": 16.582900519376746,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 41507
    },
    {
      "epoch": 16.583300039952057,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 41508
    },
    {
      "epoch": 16.583699560527368,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 41509
    },
    {
      "epoch": 16.58409908110268,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 41510
    },
    {
      "epoch": 16.584498601677986,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 41511
    },
    {
      "epoch": 16.584898122253296,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 41512
    },
    {
      "epoch": 16.585297642828607,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 41513
    },
    {
      "epoch": 16.585697163403914,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 41514
    },
    {
      "epoch": 16.586096683979225,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 41515
    },
    {
      "epoch": 16.586496204554535,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 41516
    },
    {
      "epoch": 16.586895725129843,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 41517
    },
    {
      "epoch": 16.587295245705153,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 41518
    },
    {
      "epoch": 16.587694766280464,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 41519
    },
    {
      "epoch": 16.588094286855775,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 41520
    },
    {
      "epoch": 16.58849380743108,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 41521
    },
    {
      "epoch": 16.588893328006392,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 41522
    },
    {
      "epoch": 16.589292848581703,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 41523
    },
    {
      "epoch": 16.58969236915701,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 41524
    },
    {
      "epoch": 16.59009188973232,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 41525
    },
    {
      "epoch": 16.59049141030763,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 41526
    },
    {
      "epoch": 16.590890930882942,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 41527
    },
    {
      "epoch": 16.59129045145825,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 41528
    },
    {
      "epoch": 16.59168997203356,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 41529
    },
    {
      "epoch": 16.59208949260887,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 41530
    },
    {
      "epoch": 16.592489013184178,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 41531
    },
    {
      "epoch": 16.59288853375949,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 41532
    },
    {
      "epoch": 16.5932880543348,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 41533
    },
    {
      "epoch": 16.593687574910106,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 41534
    },
    {
      "epoch": 16.594087095485417,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 41535
    },
    {
      "epoch": 16.594486616060728,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0089,
      "step": 41536
    },
    {
      "epoch": 16.59488613663604,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 41537
    },
    {
      "epoch": 16.595285657211345,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 41538
    },
    {
      "epoch": 16.595685177786656,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 41539
    },
    {
      "epoch": 16.596084698361967,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 41540
    },
    {
      "epoch": 16.596484218937274,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 41541
    },
    {
      "epoch": 16.596883739512585,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 41542
    },
    {
      "epoch": 16.597283260087895,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 41543
    },
    {
      "epoch": 16.597682780663206,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 41544
    },
    {
      "epoch": 16.598082301238513,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 41545
    },
    {
      "epoch": 16.598481821813824,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0122,
      "step": 41546
    },
    {
      "epoch": 16.598881342389134,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 41547
    },
    {
      "epoch": 16.59928086296444,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 41548
    },
    {
      "epoch": 16.599680383539752,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 41549
    },
    {
      "epoch": 16.600079904115063,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 41550
    },
    {
      "epoch": 16.60047942469037,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 41551
    },
    {
      "epoch": 16.60087894526568,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 41552
    },
    {
      "epoch": 16.60127846584099,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 41553
    },
    {
      "epoch": 16.601677986416302,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 41554
    },
    {
      "epoch": 16.60207750699161,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 41555
    },
    {
      "epoch": 16.60247702756692,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 41556
    },
    {
      "epoch": 16.60287654814223,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 41557
    },
    {
      "epoch": 16.603276068717538,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 41558
    },
    {
      "epoch": 16.60367558929285,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 41559
    },
    {
      "epoch": 16.60407510986816,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 41560
    },
    {
      "epoch": 16.60447463044347,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 41561
    },
    {
      "epoch": 16.604874151018777,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 41562
    },
    {
      "epoch": 16.605273671594087,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 41563
    },
    {
      "epoch": 16.605673192169398,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 41564
    },
    {
      "epoch": 16.606072712744705,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 41565
    },
    {
      "epoch": 16.606472233320016,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 41566
    },
    {
      "epoch": 16.606871753895327,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 41567
    },
    {
      "epoch": 16.607271274470634,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 41568
    },
    {
      "epoch": 16.607670795045944,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 41569
    },
    {
      "epoch": 16.608070315621255,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 41570
    },
    {
      "epoch": 16.608469836196566,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 41571
    },
    {
      "epoch": 16.608869356771873,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 41572
    },
    {
      "epoch": 16.609268877347183,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 41573
    },
    {
      "epoch": 16.609668397922494,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 41574
    },
    {
      "epoch": 16.6100679184978,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 41575
    },
    {
      "epoch": 16.610467439073112,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 41576
    },
    {
      "epoch": 16.610866959648423,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 41577
    },
    {
      "epoch": 16.611266480223733,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 41578
    },
    {
      "epoch": 16.61166600079904,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 41579
    },
    {
      "epoch": 16.61206552137435,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 41580
    },
    {
      "epoch": 16.612465041949662,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0087,
      "step": 41581
    },
    {
      "epoch": 16.61286456252497,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 41582
    },
    {
      "epoch": 16.61326408310028,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 41583
    },
    {
      "epoch": 16.61366360367559,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 41584
    },
    {
      "epoch": 16.614063124250897,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 41585
    },
    {
      "epoch": 16.614462644826208,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 41586
    },
    {
      "epoch": 16.61486216540152,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 41587
    },
    {
      "epoch": 16.61526168597683,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 41588
    },
    {
      "epoch": 16.615661206552137,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 41589
    },
    {
      "epoch": 16.616060727127447,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 41590
    },
    {
      "epoch": 16.616460247702758,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 41591
    },
    {
      "epoch": 16.616859768278065,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 41592
    },
    {
      "epoch": 16.617259288853376,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 41593
    },
    {
      "epoch": 16.617658809428686,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 41594
    },
    {
      "epoch": 16.618058330003997,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 41595
    },
    {
      "epoch": 16.618457850579304,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 41596
    },
    {
      "epoch": 16.618857371154615,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 41597
    },
    {
      "epoch": 16.619256891729925,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 41598
    },
    {
      "epoch": 16.619656412305233,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 41599
    },
    {
      "epoch": 16.620055932880543,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 41600
    },
    {
      "epoch": 16.620455453455854,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 41601
    },
    {
      "epoch": 16.62085497403116,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 41602
    },
    {
      "epoch": 16.62125449460647,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 41603
    },
    {
      "epoch": 16.621654015181782,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 41604
    },
    {
      "epoch": 16.622053535757093,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 41605
    },
    {
      "epoch": 16.6224530563324,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 41606
    },
    {
      "epoch": 16.62285257690771,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 41607
    },
    {
      "epoch": 16.62325209748302,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 41608
    },
    {
      "epoch": 16.62365161805833,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 41609
    },
    {
      "epoch": 16.62405113863364,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 41610
    },
    {
      "epoch": 16.62445065920895,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 41611
    },
    {
      "epoch": 16.62485017978426,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 41612
    },
    {
      "epoch": 16.625249700359568,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 41613
    },
    {
      "epoch": 16.62564922093488,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 41614
    },
    {
      "epoch": 16.62604874151019,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 41615
    },
    {
      "epoch": 16.626448262085496,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 41616
    },
    {
      "epoch": 16.626847782660807,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 41617
    },
    {
      "epoch": 16.627247303236118,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 41618
    },
    {
      "epoch": 16.627646823811425,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 41619
    },
    {
      "epoch": 16.628046344386735,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 41620
    },
    {
      "epoch": 16.628445864962046,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 41621
    },
    {
      "epoch": 16.628845385537357,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 41622
    },
    {
      "epoch": 16.629244906112664,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 41623
    },
    {
      "epoch": 16.629644426687975,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 41624
    },
    {
      "epoch": 16.630043947263285,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 41625
    },
    {
      "epoch": 16.630443467838592,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 41626
    },
    {
      "epoch": 16.630842988413903,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 41627
    },
    {
      "epoch": 16.631242508989214,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 41628
    },
    {
      "epoch": 16.63164202956452,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 41629
    },
    {
      "epoch": 16.63204155013983,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 41630
    },
    {
      "epoch": 16.632441070715142,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 41631
    },
    {
      "epoch": 16.632840591290453,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 41632
    },
    {
      "epoch": 16.63324011186576,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 41633
    },
    {
      "epoch": 16.63363963244107,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 41634
    },
    {
      "epoch": 16.63403915301638,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 41635
    },
    {
      "epoch": 16.63443867359169,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 41636
    },
    {
      "epoch": 16.634838194167,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 41637
    },
    {
      "epoch": 16.63523771474231,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 41638
    },
    {
      "epoch": 16.63563723531762,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 41639
    },
    {
      "epoch": 16.636036755892928,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 41640
    },
    {
      "epoch": 16.63643627646824,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 41641
    },
    {
      "epoch": 16.63683579704355,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 41642
    },
    {
      "epoch": 16.637235317618856,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 41643
    },
    {
      "epoch": 16.637634838194167,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 41644
    },
    {
      "epoch": 16.638034358769477,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 41645
    },
    {
      "epoch": 16.638433879344785,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 41646
    },
    {
      "epoch": 16.638833399920095,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 41647
    },
    {
      "epoch": 16.639232920495406,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 41648
    },
    {
      "epoch": 16.639632441070717,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 41649
    },
    {
      "epoch": 16.640031961646024,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 41650
    },
    {
      "epoch": 16.640431482221334,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0078,
      "step": 41651
    },
    {
      "epoch": 16.640831002796645,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 41652
    },
    {
      "epoch": 16.641230523371952,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 41653
    },
    {
      "epoch": 16.641630043947263,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 41654
    },
    {
      "epoch": 16.642029564522574,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 41655
    },
    {
      "epoch": 16.642429085097884,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 41656
    },
    {
      "epoch": 16.64282860567319,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 41657
    },
    {
      "epoch": 16.643228126248502,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 41658
    },
    {
      "epoch": 16.643627646823813,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 41659
    },
    {
      "epoch": 16.64402716739912,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 41660
    },
    {
      "epoch": 16.64442668797443,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 41661
    },
    {
      "epoch": 16.64482620854974,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 41662
    },
    {
      "epoch": 16.64522572912505,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 41663
    },
    {
      "epoch": 16.64562524970036,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 41664
    },
    {
      "epoch": 16.64602477027567,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 41665
    },
    {
      "epoch": 16.64642429085098,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 41666
    },
    {
      "epoch": 16.646823811426287,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 41667
    },
    {
      "epoch": 16.647223332001598,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 41668
    },
    {
      "epoch": 16.64762285257691,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 41669
    },
    {
      "epoch": 16.648022373152216,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 41670
    },
    {
      "epoch": 16.648421893727527,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 41671
    },
    {
      "epoch": 16.648821414302837,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 41672
    },
    {
      "epoch": 16.649220934878148,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 41673
    },
    {
      "epoch": 16.649620455453455,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 41674
    },
    {
      "epoch": 16.650019976028766,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 41675
    },
    {
      "epoch": 16.650419496604076,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 41676
    },
    {
      "epoch": 16.650819017179384,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 41677
    },
    {
      "epoch": 16.651218537754694,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 41678
    },
    {
      "epoch": 16.651618058330005,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 41679
    },
    {
      "epoch": 16.652017578905312,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 41680
    },
    {
      "epoch": 16.652417099480623,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 41681
    },
    {
      "epoch": 16.652816620055933,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 41682
    },
    {
      "epoch": 16.653216140631244,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 41683
    },
    {
      "epoch": 16.65361566120655,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 41684
    },
    {
      "epoch": 16.654015181781862,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 41685
    },
    {
      "epoch": 16.654414702357172,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 41686
    },
    {
      "epoch": 16.65481422293248,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 41687
    },
    {
      "epoch": 16.65521374350779,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 41688
    },
    {
      "epoch": 16.6556132640831,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 41689
    },
    {
      "epoch": 16.65601278465841,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 41690
    },
    {
      "epoch": 16.65641230523372,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 41691
    },
    {
      "epoch": 16.65681182580903,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0126,
      "step": 41692
    },
    {
      "epoch": 16.65721134638434,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 41693
    },
    {
      "epoch": 16.657610866959647,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 41694
    },
    {
      "epoch": 16.658010387534958,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 41695
    },
    {
      "epoch": 16.65840990811027,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 41696
    },
    {
      "epoch": 16.658809428685576,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 41697
    },
    {
      "epoch": 16.659208949260886,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 41698
    },
    {
      "epoch": 16.659608469836197,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 41699
    },
    {
      "epoch": 16.660007990411508,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 41700
    },
    {
      "epoch": 16.660407510986815,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 41701
    },
    {
      "epoch": 16.660807031562126,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 41702
    },
    {
      "epoch": 16.661206552137436,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 41703
    },
    {
      "epoch": 16.661606072712743,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 41704
    },
    {
      "epoch": 16.662005593288054,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 41705
    },
    {
      "epoch": 16.662405113863365,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 41706
    },
    {
      "epoch": 16.66280463443867,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 41707
    },
    {
      "epoch": 16.663204155013982,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 41708
    },
    {
      "epoch": 16.663603675589293,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 41709
    },
    {
      "epoch": 16.664003196164604,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 41710
    },
    {
      "epoch": 16.66440271673991,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.009,
      "step": 41711
    },
    {
      "epoch": 16.66480223731522,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 41712
    },
    {
      "epoch": 16.665201757890532,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 41713
    },
    {
      "epoch": 16.66560127846584,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 41714
    },
    {
      "epoch": 16.66600079904115,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 41715
    },
    {
      "epoch": 16.66640031961646,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 41716
    },
    {
      "epoch": 16.66679984019177,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 41717
    },
    {
      "epoch": 16.66719936076708,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0134,
      "step": 41718
    },
    {
      "epoch": 16.66759888134239,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 41719
    },
    {
      "epoch": 16.6679984019177,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 41720
    },
    {
      "epoch": 16.668397922493007,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 41721
    },
    {
      "epoch": 16.668797443068318,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 41722
    },
    {
      "epoch": 16.66919696364363,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 41723
    },
    {
      "epoch": 16.669596484218935,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 41724
    },
    {
      "epoch": 16.669996004794246,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 41725
    },
    {
      "epoch": 16.670395525369557,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 41726
    },
    {
      "epoch": 16.670795045944867,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 41727
    },
    {
      "epoch": 16.671194566520175,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 41728
    },
    {
      "epoch": 16.671594087095485,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 41729
    },
    {
      "epoch": 16.671993607670796,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 41730
    },
    {
      "epoch": 16.672393128246103,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0434,
      "step": 41731
    },
    {
      "epoch": 16.672792648821414,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 41732
    },
    {
      "epoch": 16.673192169396724,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 41733
    },
    {
      "epoch": 16.673591689972035,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 41734
    },
    {
      "epoch": 16.673991210547342,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 41735
    },
    {
      "epoch": 16.674390731122653,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 41736
    },
    {
      "epoch": 16.674790251697964,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 41737
    },
    {
      "epoch": 16.67518977227327,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 41738
    },
    {
      "epoch": 16.67558929284858,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 41739
    },
    {
      "epoch": 16.675988813423892,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 41740
    },
    {
      "epoch": 16.6763883339992,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 41741
    },
    {
      "epoch": 16.67678785457451,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 41742
    },
    {
      "epoch": 16.67718737514982,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 41743
    },
    {
      "epoch": 16.67758689572513,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 41744
    },
    {
      "epoch": 16.67798641630044,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 41745
    },
    {
      "epoch": 16.67838593687575,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.006,
      "step": 41746
    },
    {
      "epoch": 16.67878545745106,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 41747
    },
    {
      "epoch": 16.679184978026367,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 41748
    },
    {
      "epoch": 16.679584498601677,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 41749
    },
    {
      "epoch": 16.679984019176988,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 41750
    },
    {
      "epoch": 16.6803835397523,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0114,
      "step": 41751
    },
    {
      "epoch": 16.680783060327606,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 41752
    },
    {
      "epoch": 16.681182580902917,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 41753
    },
    {
      "epoch": 16.681582101478227,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 41754
    },
    {
      "epoch": 16.681981622053534,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 41755
    },
    {
      "epoch": 16.682381142628845,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 41756
    },
    {
      "epoch": 16.682780663204156,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 41757
    },
    {
      "epoch": 16.683180183779463,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 41758
    },
    {
      "epoch": 16.683579704354774,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 41759
    },
    {
      "epoch": 16.683979224930084,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 41760
    },
    {
      "epoch": 16.684378745505395,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 41761
    },
    {
      "epoch": 16.684778266080702,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 41762
    },
    {
      "epoch": 16.685177786656013,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 41763
    },
    {
      "epoch": 16.685577307231323,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 41764
    },
    {
      "epoch": 16.68597682780663,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 41765
    },
    {
      "epoch": 16.68637634838194,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 41766
    },
    {
      "epoch": 16.686775868957252,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 41767
    },
    {
      "epoch": 16.687175389532563,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 41768
    },
    {
      "epoch": 16.68757491010787,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 41769
    },
    {
      "epoch": 16.68797443068318,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 41770
    },
    {
      "epoch": 16.68837395125849,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 41771
    },
    {
      "epoch": 16.688773471833798,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 41772
    },
    {
      "epoch": 16.68917299240911,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 41773
    },
    {
      "epoch": 16.68957251298442,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 41774
    },
    {
      "epoch": 16.689972033559727,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 41775
    },
    {
      "epoch": 16.690371554135037,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 41776
    },
    {
      "epoch": 16.690771074710348,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 41777
    },
    {
      "epoch": 16.69117059528566,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 41778
    },
    {
      "epoch": 16.691570115860966,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 41779
    },
    {
      "epoch": 16.691969636436276,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 41780
    },
    {
      "epoch": 16.692369157011587,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 41781
    },
    {
      "epoch": 16.692768677586894,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 41782
    },
    {
      "epoch": 16.693168198162205,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 41783
    },
    {
      "epoch": 16.693567718737516,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 41784
    },
    {
      "epoch": 16.693967239312826,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 41785
    },
    {
      "epoch": 16.694366759888133,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 41786
    },
    {
      "epoch": 16.694766280463444,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 41787
    },
    {
      "epoch": 16.695165801038755,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 41788
    },
    {
      "epoch": 16.695565321614062,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 41789
    },
    {
      "epoch": 16.695964842189372,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 41790
    },
    {
      "epoch": 16.696364362764683,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 41791
    },
    {
      "epoch": 16.69676388333999,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 41792
    },
    {
      "epoch": 16.6971634039153,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 41793
    },
    {
      "epoch": 16.69756292449061,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 41794
    },
    {
      "epoch": 16.697962445065922,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 41795
    },
    {
      "epoch": 16.69836196564123,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 41796
    },
    {
      "epoch": 16.69876148621654,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 41797
    },
    {
      "epoch": 16.69916100679185,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 41798
    },
    {
      "epoch": 16.699560527367158,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 41799
    },
    {
      "epoch": 16.69996004794247,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 41800
    },
    {
      "epoch": 16.70035956851778,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 41801
    },
    {
      "epoch": 16.70075908909309,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 41802
    },
    {
      "epoch": 16.701158609668397,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 41803
    },
    {
      "epoch": 16.701558130243708,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 41804
    },
    {
      "epoch": 16.70195765081902,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 41805
    },
    {
      "epoch": 16.702357171394326,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 41806
    },
    {
      "epoch": 16.702756691969636,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 41807
    },
    {
      "epoch": 16.703156212544947,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 41808
    },
    {
      "epoch": 16.703555733120254,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 41809
    },
    {
      "epoch": 16.703955253695565,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 41810
    },
    {
      "epoch": 16.704354774270875,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 41811
    },
    {
      "epoch": 16.704754294846186,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.01,
      "step": 41812
    },
    {
      "epoch": 16.705153815421493,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 41813
    },
    {
      "epoch": 16.705553335996804,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 41814
    },
    {
      "epoch": 16.705952856572114,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 41815
    },
    {
      "epoch": 16.70635237714742,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 41816
    },
    {
      "epoch": 16.706751897722732,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 41817
    },
    {
      "epoch": 16.707151418298043,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 41818
    },
    {
      "epoch": 16.707550938873354,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 41819
    },
    {
      "epoch": 16.70795045944866,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 41820
    },
    {
      "epoch": 16.70834998002397,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 41821
    },
    {
      "epoch": 16.708749500599282,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 41822
    },
    {
      "epoch": 16.70914902117459,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 41823
    },
    {
      "epoch": 16.7095485417499,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 41824
    },
    {
      "epoch": 16.70994806232521,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 41825
    },
    {
      "epoch": 16.710347582900518,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0095,
      "step": 41826
    },
    {
      "epoch": 16.71074710347583,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 41827
    },
    {
      "epoch": 16.71114662405114,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 41828
    },
    {
      "epoch": 16.71154614462645,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 41829
    },
    {
      "epoch": 16.711945665201757,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 41830
    },
    {
      "epoch": 16.712345185777068,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 41831
    },
    {
      "epoch": 16.712744706352378,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 41832
    },
    {
      "epoch": 16.713144226927685,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 41833
    },
    {
      "epoch": 16.713543747502996,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 41834
    },
    {
      "epoch": 16.713943268078307,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0105,
      "step": 41835
    },
    {
      "epoch": 16.714342788653617,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 41836
    },
    {
      "epoch": 16.714742309228924,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 41837
    },
    {
      "epoch": 16.715141829804235,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 41838
    },
    {
      "epoch": 16.715541350379546,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 41839
    },
    {
      "epoch": 16.715940870954853,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0457,
      "step": 41840
    },
    {
      "epoch": 16.716340391530164,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 41841
    },
    {
      "epoch": 16.716739912105474,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 41842
    },
    {
      "epoch": 16.71713943268078,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 41843
    },
    {
      "epoch": 16.717538953256092,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 41844
    },
    {
      "epoch": 16.717938473831403,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 41845
    },
    {
      "epoch": 16.718337994406713,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 41846
    },
    {
      "epoch": 16.71873751498202,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 41847
    },
    {
      "epoch": 16.71913703555733,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 41848
    },
    {
      "epoch": 16.719536556132642,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0425,
      "step": 41849
    },
    {
      "epoch": 16.71993607670795,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0442,
      "step": 41850
    },
    {
      "epoch": 16.72033559728326,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 41851
    },
    {
      "epoch": 16.72073511785857,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 41852
    },
    {
      "epoch": 16.72113463843388,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 41853
    },
    {
      "epoch": 16.721534159009188,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 41854
    },
    {
      "epoch": 16.7219336795845,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 41855
    },
    {
      "epoch": 16.72233320015981,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0091,
      "step": 41856
    },
    {
      "epoch": 16.722732720735117,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 41857
    },
    {
      "epoch": 16.723132241310427,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 41858
    },
    {
      "epoch": 16.723531761885738,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 41859
    },
    {
      "epoch": 16.723931282461045,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 41860
    },
    {
      "epoch": 16.724330803036356,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 41861
    },
    {
      "epoch": 16.724730323611666,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 41862
    },
    {
      "epoch": 16.725129844186977,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 41863
    },
    {
      "epoch": 16.725529364762284,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 41864
    },
    {
      "epoch": 16.725928885337595,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 41865
    },
    {
      "epoch": 16.726328405912906,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 41866
    },
    {
      "epoch": 16.726727926488213,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 41867
    },
    {
      "epoch": 16.727127447063523,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 41868
    },
    {
      "epoch": 16.727526967638834,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 41869
    },
    {
      "epoch": 16.727926488214145,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 41870
    },
    {
      "epoch": 16.728326008789452,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 41871
    },
    {
      "epoch": 16.728725529364763,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 41872
    },
    {
      "epoch": 16.729125049940073,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 41873
    },
    {
      "epoch": 16.72952457051538,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 41874
    },
    {
      "epoch": 16.72992409109069,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 41875
    },
    {
      "epoch": 16.730323611666,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 41876
    },
    {
      "epoch": 16.73072313224131,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 41877
    },
    {
      "epoch": 16.73112265281662,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 41878
    },
    {
      "epoch": 16.73152217339193,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 41879
    },
    {
      "epoch": 16.73192169396724,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 41880
    },
    {
      "epoch": 16.732321214542548,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 41881
    },
    {
      "epoch": 16.73272073511786,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 41882
    },
    {
      "epoch": 16.73312025569317,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 41883
    },
    {
      "epoch": 16.733519776268476,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 41884
    },
    {
      "epoch": 16.733919296843787,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 41885
    },
    {
      "epoch": 16.734318817419098,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 41886
    },
    {
      "epoch": 16.73471833799441,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 41887
    },
    {
      "epoch": 16.735117858569716,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 41888
    },
    {
      "epoch": 16.735517379145026,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 41889
    },
    {
      "epoch": 16.735916899720337,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 41890
    },
    {
      "epoch": 16.736316420295644,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 41891
    },
    {
      "epoch": 16.736715940870955,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 41892
    },
    {
      "epoch": 16.737115461446265,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 41893
    },
    {
      "epoch": 16.737514982021573,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 41894
    },
    {
      "epoch": 16.737914502596883,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 41895
    },
    {
      "epoch": 16.738314023172194,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 41896
    },
    {
      "epoch": 16.738713543747505,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 41897
    },
    {
      "epoch": 16.73911306432281,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 41898
    },
    {
      "epoch": 16.739512584898122,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 41899
    },
    {
      "epoch": 16.739912105473433,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 41900
    },
    {
      "epoch": 16.74031162604874,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 41901
    },
    {
      "epoch": 16.74071114662405,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 41902
    },
    {
      "epoch": 16.74111066719936,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 41903
    },
    {
      "epoch": 16.741510187774672,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 41904
    },
    {
      "epoch": 16.74190970834998,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 41905
    },
    {
      "epoch": 16.74230922892529,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 41906
    },
    {
      "epoch": 16.7427087495006,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 41907
    },
    {
      "epoch": 16.743108270075908,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 41908
    },
    {
      "epoch": 16.74350779065122,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 41909
    },
    {
      "epoch": 16.74390731122653,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 41910
    },
    {
      "epoch": 16.744306831801836,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 41911
    },
    {
      "epoch": 16.744706352377147,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 41912
    },
    {
      "epoch": 16.745105872952458,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 41913
    },
    {
      "epoch": 16.74550539352777,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 41914
    },
    {
      "epoch": 16.745904914103075,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 41915
    },
    {
      "epoch": 16.746304434678386,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 41916
    },
    {
      "epoch": 16.746703955253697,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 41917
    },
    {
      "epoch": 16.747103475829004,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 41918
    },
    {
      "epoch": 16.747502996404315,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 41919
    },
    {
      "epoch": 16.747902516979625,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 41920
    },
    {
      "epoch": 16.748302037554936,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 41921
    },
    {
      "epoch": 16.748701558130243,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 41922
    },
    {
      "epoch": 16.749101078705554,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 41923
    },
    {
      "epoch": 16.749500599280864,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 41924
    },
    {
      "epoch": 16.74990011985617,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 41925
    },
    {
      "epoch": 16.750299640431482,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 41926
    },
    {
      "epoch": 16.750699161006793,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 41927
    },
    {
      "epoch": 16.7510986815821,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 41928
    },
    {
      "epoch": 16.75149820215741,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 41929
    },
    {
      "epoch": 16.75189772273272,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 41930
    },
    {
      "epoch": 16.752297243308032,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 41931
    },
    {
      "epoch": 16.75269676388334,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 41932
    },
    {
      "epoch": 16.75309628445865,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 41933
    },
    {
      "epoch": 16.75349580503396,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 41934
    },
    {
      "epoch": 16.753895325609268,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 41935
    },
    {
      "epoch": 16.754294846184578,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 41936
    },
    {
      "epoch": 16.75469436675989,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 41937
    },
    {
      "epoch": 16.7550938873352,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 41938
    },
    {
      "epoch": 16.755493407910507,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 41939
    },
    {
      "epoch": 16.755892928485817,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 41940
    },
    {
      "epoch": 16.756292449061128,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 41941
    },
    {
      "epoch": 16.756691969636435,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 41942
    },
    {
      "epoch": 16.757091490211746,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 41943
    },
    {
      "epoch": 16.757491010787056,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 41944
    },
    {
      "epoch": 16.757890531362364,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 41945
    },
    {
      "epoch": 16.758290051937674,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0451,
      "step": 41946
    },
    {
      "epoch": 16.758689572512985,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 41947
    },
    {
      "epoch": 16.759089093088296,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 41948
    },
    {
      "epoch": 16.759488613663603,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 41949
    },
    {
      "epoch": 16.759888134238913,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 41950
    },
    {
      "epoch": 16.760287654814224,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 41951
    },
    {
      "epoch": 16.76068717538953,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 41952
    },
    {
      "epoch": 16.761086695964842,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 41953
    },
    {
      "epoch": 16.761486216540153,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 41954
    },
    {
      "epoch": 16.761885737115463,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 41955
    },
    {
      "epoch": 16.76228525769077,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 41956
    },
    {
      "epoch": 16.76268477826608,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 41957
    },
    {
      "epoch": 16.76308429884139,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 41958
    },
    {
      "epoch": 16.7634838194167,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 41959
    },
    {
      "epoch": 16.76388333999201,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 41960
    },
    {
      "epoch": 16.76428286056732,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 41961
    },
    {
      "epoch": 16.764682381142627,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 41962
    },
    {
      "epoch": 16.765081901717938,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 41963
    },
    {
      "epoch": 16.76548142229325,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 41964
    },
    {
      "epoch": 16.76588094286856,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 41965
    },
    {
      "epoch": 16.766280463443866,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 41966
    },
    {
      "epoch": 16.766679984019177,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 41967
    },
    {
      "epoch": 16.767079504594488,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 41968
    },
    {
      "epoch": 16.767479025169795,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 41969
    },
    {
      "epoch": 16.767878545745106,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 41970
    },
    {
      "epoch": 16.768278066320416,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 41971
    },
    {
      "epoch": 16.768677586895727,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 41972
    },
    {
      "epoch": 16.769077107471034,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 41973
    },
    {
      "epoch": 16.769476628046345,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 41974
    },
    {
      "epoch": 16.769876148621655,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 41975
    },
    {
      "epoch": 16.770275669196963,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 41976
    },
    {
      "epoch": 16.770675189772273,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 41977
    },
    {
      "epoch": 16.771074710347584,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 41978
    },
    {
      "epoch": 16.77147423092289,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 41979
    },
    {
      "epoch": 16.7718737514982,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 41980
    },
    {
      "epoch": 16.772273272073512,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 41981
    },
    {
      "epoch": 16.772672792648823,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 41982
    },
    {
      "epoch": 16.77307231322413,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 41983
    },
    {
      "epoch": 16.77347183379944,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 41984
    },
    {
      "epoch": 16.77387135437475,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 41985
    },
    {
      "epoch": 16.77427087495006,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 41986
    },
    {
      "epoch": 16.77467039552537,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 41987
    },
    {
      "epoch": 16.77506991610068,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 41988
    },
    {
      "epoch": 16.775469436675987,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 41989
    },
    {
      "epoch": 16.775868957251298,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 41990
    },
    {
      "epoch": 16.77626847782661,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 41991
    },
    {
      "epoch": 16.77666799840192,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 41992
    },
    {
      "epoch": 16.777067518977226,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 41993
    },
    {
      "epoch": 16.777467039552537,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 41994
    },
    {
      "epoch": 16.777866560127848,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 41995
    },
    {
      "epoch": 16.778266080703155,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 41996
    },
    {
      "epoch": 16.778665601278465,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 41997
    },
    {
      "epoch": 16.779065121853776,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 41998
    },
    {
      "epoch": 16.779464642429087,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 41999
    },
    {
      "epoch": 16.779864163004394,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 42000
    },
    {
      "epoch": 16.780263683579705,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 42001
    },
    {
      "epoch": 16.780663204155015,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 42002
    },
    {
      "epoch": 16.781062724730322,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 42003
    },
    {
      "epoch": 16.781462245305633,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 42004
    },
    {
      "epoch": 16.781861765880944,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 42005
    },
    {
      "epoch": 16.78226128645625,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 42006
    },
    {
      "epoch": 16.78266080703156,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 42007
    },
    {
      "epoch": 16.783060327606872,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 42008
    },
    {
      "epoch": 16.783459848182183,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 42009
    },
    {
      "epoch": 16.78385936875749,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 42010
    },
    {
      "epoch": 16.7842588893328,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 42011
    },
    {
      "epoch": 16.78465840990811,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0064,
      "step": 42012
    },
    {
      "epoch": 16.78505793048342,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 42013
    },
    {
      "epoch": 16.78545745105873,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 42014
    },
    {
      "epoch": 16.78585697163404,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 42015
    },
    {
      "epoch": 16.78625649220935,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 42016
    },
    {
      "epoch": 16.786656012784658,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 42017
    },
    {
      "epoch": 16.78705553335997,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 42018
    },
    {
      "epoch": 16.78745505393528,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 42019
    },
    {
      "epoch": 16.787854574510586,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 42020
    },
    {
      "epoch": 16.788254095085897,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 42021
    },
    {
      "epoch": 16.788653615661207,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 42022
    },
    {
      "epoch": 16.789053136236515,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 42023
    },
    {
      "epoch": 16.789452656811825,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 42024
    },
    {
      "epoch": 16.789852177387136,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 42025
    },
    {
      "epoch": 16.790251697962447,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 42026
    },
    {
      "epoch": 16.790651218537754,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 42027
    },
    {
      "epoch": 16.791050739113064,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 42028
    },
    {
      "epoch": 16.791450259688375,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 42029
    },
    {
      "epoch": 16.791849780263682,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 42030
    },
    {
      "epoch": 16.792249300838993,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 42031
    },
    {
      "epoch": 16.792648821414303,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 42032
    },
    {
      "epoch": 16.793048341989614,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 42033
    },
    {
      "epoch": 16.79344786256492,
      "grad_norm": 0.171875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 42034
    },
    {
      "epoch": 16.793847383140232,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 42035
    },
    {
      "epoch": 16.794246903715543,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 42036
    },
    {
      "epoch": 16.79464642429085,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 42037
    },
    {
      "epoch": 16.79504594486616,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 42038
    },
    {
      "epoch": 16.79544546544147,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 42039
    },
    {
      "epoch": 16.795844986016778,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 42040
    },
    {
      "epoch": 16.79624450659209,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 42041
    },
    {
      "epoch": 16.7966440271674,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 42042
    },
    {
      "epoch": 16.79704354774271,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0134,
      "step": 42043
    },
    {
      "epoch": 16.797443068318017,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 42044
    },
    {
      "epoch": 16.797842588893328,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 42045
    },
    {
      "epoch": 16.79824210946864,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 42046
    },
    {
      "epoch": 16.798641630043946,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 42047
    },
    {
      "epoch": 16.799041150619257,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 42048
    },
    {
      "epoch": 16.799440671194567,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 42049
    },
    {
      "epoch": 16.799840191769878,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 42050
    },
    {
      "epoch": 16.800239712345185,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 42051
    },
    {
      "epoch": 16.800639232920496,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 42052
    },
    {
      "epoch": 16.801038753495806,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 42053
    },
    {
      "epoch": 16.801438274071113,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 42054
    },
    {
      "epoch": 16.801837794646424,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 42055
    },
    {
      "epoch": 16.802237315221735,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 42056
    },
    {
      "epoch": 16.802636835797042,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 42057
    },
    {
      "epoch": 16.803036356372353,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 42058
    },
    {
      "epoch": 16.803435876947663,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 42059
    },
    {
      "epoch": 16.803835397522974,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 42060
    },
    {
      "epoch": 16.80423491809828,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 42061
    },
    {
      "epoch": 16.80463443867359,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 42062
    },
    {
      "epoch": 16.805033959248902,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 42063
    },
    {
      "epoch": 16.80543347982421,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 42064
    },
    {
      "epoch": 16.80583300039952,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 42065
    },
    {
      "epoch": 16.80623252097483,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 42066
    },
    {
      "epoch": 16.806632041550138,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 42067
    },
    {
      "epoch": 16.80703156212545,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 42068
    },
    {
      "epoch": 16.80743108270076,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 42069
    },
    {
      "epoch": 16.80783060327607,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 42070
    },
    {
      "epoch": 16.808230123851377,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 42071
    },
    {
      "epoch": 16.808629644426688,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 42072
    },
    {
      "epoch": 16.809029165002,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 42073
    },
    {
      "epoch": 16.809428685577306,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 42074
    },
    {
      "epoch": 16.809828206152616,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 42075
    },
    {
      "epoch": 16.810227726727927,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 42076
    },
    {
      "epoch": 16.810627247303238,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 42077
    },
    {
      "epoch": 16.811026767878545,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 42078
    },
    {
      "epoch": 16.811426288453855,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 42079
    },
    {
      "epoch": 16.811825809029166,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 42080
    },
    {
      "epoch": 16.812225329604473,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 42081
    },
    {
      "epoch": 16.812624850179784,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 42082
    },
    {
      "epoch": 16.813024370755095,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 42083
    },
    {
      "epoch": 16.8134238913304,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 42084
    },
    {
      "epoch": 16.813823411905712,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 42085
    },
    {
      "epoch": 16.814222932481023,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 42086
    },
    {
      "epoch": 16.814622453056334,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 42087
    },
    {
      "epoch": 16.81502197363164,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 42088
    },
    {
      "epoch": 16.81542149420695,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 42089
    },
    {
      "epoch": 16.815821014782262,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 42090
    },
    {
      "epoch": 16.81622053535757,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 42091
    },
    {
      "epoch": 16.81662005593288,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 42092
    },
    {
      "epoch": 16.81701957650819,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 42093
    },
    {
      "epoch": 16.8174190970835,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 42094
    },
    {
      "epoch": 16.81781861765881,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 42095
    },
    {
      "epoch": 16.81821813823412,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 42096
    },
    {
      "epoch": 16.81861765880943,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 42097
    },
    {
      "epoch": 16.819017179384737,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 42098
    },
    {
      "epoch": 16.819416699960048,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 42099
    },
    {
      "epoch": 16.81981622053536,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 42100
    },
    {
      "epoch": 16.820215741110665,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 42101
    },
    {
      "epoch": 16.820615261685976,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 42102
    },
    {
      "epoch": 16.821014782261287,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 42103
    },
    {
      "epoch": 16.821414302836597,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 42104
    },
    {
      "epoch": 16.821813823411905,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 42105
    },
    {
      "epoch": 16.822213343987215,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 42106
    },
    {
      "epoch": 16.822612864562526,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 42107
    },
    {
      "epoch": 16.823012385137833,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 42108
    },
    {
      "epoch": 16.823411905713144,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 42109
    },
    {
      "epoch": 16.823811426288454,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 42110
    },
    {
      "epoch": 16.824210946863765,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 42111
    },
    {
      "epoch": 16.824610467439072,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 42112
    },
    {
      "epoch": 16.825009988014383,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 42113
    },
    {
      "epoch": 16.825409508589694,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 42114
    },
    {
      "epoch": 16.825809029165,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 42115
    },
    {
      "epoch": 16.82620854974031,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 42116
    },
    {
      "epoch": 16.826608070315622,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 42117
    },
    {
      "epoch": 16.82700759089093,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 42118
    },
    {
      "epoch": 16.82740711146624,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 42119
    },
    {
      "epoch": 16.82780663204155,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 42120
    },
    {
      "epoch": 16.82820615261686,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0423,
      "step": 42121
    },
    {
      "epoch": 16.82860567319217,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 42122
    },
    {
      "epoch": 16.82900519376748,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 42123
    },
    {
      "epoch": 16.82940471434279,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 42124
    },
    {
      "epoch": 16.829804234918097,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 42125
    },
    {
      "epoch": 16.830203755493407,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 42126
    },
    {
      "epoch": 16.830603276068718,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 42127
    },
    {
      "epoch": 16.83100279664403,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 42128
    },
    {
      "epoch": 16.831402317219336,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 42129
    },
    {
      "epoch": 16.831801837794647,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 42130
    },
    {
      "epoch": 16.832201358369957,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 42131
    },
    {
      "epoch": 16.832600878945264,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 42132
    },
    {
      "epoch": 16.833000399520575,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 42133
    },
    {
      "epoch": 16.833399920095886,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 42134
    },
    {
      "epoch": 16.833799440671193,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 42135
    },
    {
      "epoch": 16.834198961246504,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 42136
    },
    {
      "epoch": 16.834598481821814,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 42137
    },
    {
      "epoch": 16.834998002397125,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 42138
    },
    {
      "epoch": 16.835397522972432,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 42139
    },
    {
      "epoch": 16.835797043547743,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 42140
    },
    {
      "epoch": 16.836196564123053,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 42141
    },
    {
      "epoch": 16.83659608469836,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 42142
    },
    {
      "epoch": 16.83699560527367,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 42143
    },
    {
      "epoch": 16.837395125848982,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 42144
    },
    {
      "epoch": 16.837794646424292,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0111,
      "step": 42145
    },
    {
      "epoch": 16.8381941669996,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 42146
    },
    {
      "epoch": 16.83859368757491,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 42147
    },
    {
      "epoch": 16.83899320815022,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 42148
    },
    {
      "epoch": 16.839392728725528,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 42149
    },
    {
      "epoch": 16.83979224930084,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0115,
      "step": 42150
    },
    {
      "epoch": 16.84019176987615,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 42151
    },
    {
      "epoch": 16.840591290451457,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 42152
    },
    {
      "epoch": 16.840990811026767,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 42153
    },
    {
      "epoch": 16.841390331602078,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 42154
    },
    {
      "epoch": 16.84178985217739,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 42155
    },
    {
      "epoch": 16.842189372752696,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 42156
    },
    {
      "epoch": 16.842588893328006,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 42157
    },
    {
      "epoch": 16.842988413903317,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0096,
      "step": 42158
    },
    {
      "epoch": 16.843387934478624,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 42159
    },
    {
      "epoch": 16.843787455053935,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 42160
    },
    {
      "epoch": 16.844186975629245,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 42161
    },
    {
      "epoch": 16.844586496204556,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 42162
    },
    {
      "epoch": 16.844986016779863,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 42163
    },
    {
      "epoch": 16.845385537355174,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 42164
    },
    {
      "epoch": 16.845785057930485,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 42165
    },
    {
      "epoch": 16.84618457850579,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 42166
    },
    {
      "epoch": 16.846584099081102,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 42167
    },
    {
      "epoch": 16.846983619656413,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 42168
    },
    {
      "epoch": 16.84738314023172,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 42169
    },
    {
      "epoch": 16.84778266080703,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 42170
    },
    {
      "epoch": 16.84818218138234,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 42171
    },
    {
      "epoch": 16.848581701957652,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 42172
    },
    {
      "epoch": 16.84898122253296,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 42173
    },
    {
      "epoch": 16.84938074310827,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 42174
    },
    {
      "epoch": 16.84978026368358,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 42175
    },
    {
      "epoch": 16.850179784258888,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 42176
    },
    {
      "epoch": 16.8505793048342,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 42177
    },
    {
      "epoch": 16.85097882540951,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 42178
    },
    {
      "epoch": 16.85137834598482,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 42179
    },
    {
      "epoch": 16.851777866560127,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 42180
    },
    {
      "epoch": 16.852177387135438,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 42181
    },
    {
      "epoch": 16.85257690771075,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 42182
    },
    {
      "epoch": 16.852976428286055,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 42183
    },
    {
      "epoch": 16.853375948861366,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 42184
    },
    {
      "epoch": 16.853775469436677,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 42185
    },
    {
      "epoch": 16.854174990011984,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 42186
    },
    {
      "epoch": 16.854574510587295,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.012,
      "step": 42187
    },
    {
      "epoch": 16.854974031162605,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 42188
    },
    {
      "epoch": 16.855373551737916,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0084,
      "step": 42189
    },
    {
      "epoch": 16.855773072313223,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 42190
    },
    {
      "epoch": 16.856172592888534,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 42191
    },
    {
      "epoch": 16.856572113463844,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 42192
    },
    {
      "epoch": 16.85697163403915,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 42193
    },
    {
      "epoch": 16.857371154614462,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 42194
    },
    {
      "epoch": 16.857770675189773,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 42195
    },
    {
      "epoch": 16.858170195765084,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 42196
    },
    {
      "epoch": 16.85856971634039,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 42197
    },
    {
      "epoch": 16.8589692369157,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 42198
    },
    {
      "epoch": 16.859368757491012,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 42199
    },
    {
      "epoch": 16.85976827806632,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 42200
    },
    {
      "epoch": 16.86016779864163,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 42201
    },
    {
      "epoch": 16.86056731921694,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 42202
    },
    {
      "epoch": 16.860966839792248,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 42203
    },
    {
      "epoch": 16.86136636036756,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 42204
    },
    {
      "epoch": 16.86176588094287,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 42205
    },
    {
      "epoch": 16.86216540151818,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 42206
    },
    {
      "epoch": 16.862564922093487,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 42207
    },
    {
      "epoch": 16.862964442668797,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 42208
    },
    {
      "epoch": 16.863363963244108,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 42209
    },
    {
      "epoch": 16.863763483819415,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 42210
    },
    {
      "epoch": 16.864163004394726,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 42211
    },
    {
      "epoch": 16.864562524970037,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 42212
    },
    {
      "epoch": 16.864962045545347,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 42213
    },
    {
      "epoch": 16.865361566120654,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 42214
    },
    {
      "epoch": 16.865761086695965,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 42215
    },
    {
      "epoch": 16.866160607271276,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 42216
    },
    {
      "epoch": 16.866560127846583,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 42217
    },
    {
      "epoch": 16.866959648421894,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 42218
    },
    {
      "epoch": 16.867359168997204,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 42219
    },
    {
      "epoch": 16.86775868957251,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 42220
    },
    {
      "epoch": 16.868158210147822,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 42221
    },
    {
      "epoch": 16.868557730723133,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 42222
    },
    {
      "epoch": 16.868957251298443,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 42223
    },
    {
      "epoch": 16.86935677187375,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 42224
    },
    {
      "epoch": 16.86975629244906,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 42225
    },
    {
      "epoch": 16.870155813024372,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 42226
    },
    {
      "epoch": 16.87055533359968,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0117,
      "step": 42227
    },
    {
      "epoch": 16.87095485417499,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 42228
    },
    {
      "epoch": 16.8713543747503,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 42229
    },
    {
      "epoch": 16.87175389532561,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 42230
    },
    {
      "epoch": 16.872153415900918,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 42231
    },
    {
      "epoch": 16.87255293647623,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 42232
    },
    {
      "epoch": 16.87295245705154,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 42233
    },
    {
      "epoch": 16.873351977626847,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 42234
    },
    {
      "epoch": 16.873751498202157,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 42235
    },
    {
      "epoch": 16.874151018777468,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 42236
    },
    {
      "epoch": 16.874550539352775,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 42237
    },
    {
      "epoch": 16.874950059928086,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 42238
    },
    {
      "epoch": 16.875349580503396,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 42239
    },
    {
      "epoch": 16.875749101078707,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 42240
    },
    {
      "epoch": 16.876148621654014,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 42241
    },
    {
      "epoch": 16.876548142229325,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 42242
    },
    {
      "epoch": 16.876947662804636,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 42243
    },
    {
      "epoch": 16.877347183379943,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 42244
    },
    {
      "epoch": 16.877746703955253,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 42245
    },
    {
      "epoch": 16.878146224530564,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 42246
    },
    {
      "epoch": 16.878545745105875,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 42247
    },
    {
      "epoch": 16.878945265681182,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 42248
    },
    {
      "epoch": 16.879344786256492,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 42249
    },
    {
      "epoch": 16.879744306831803,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 42250
    },
    {
      "epoch": 16.88014382740711,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 42251
    },
    {
      "epoch": 16.88054334798242,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 42252
    },
    {
      "epoch": 16.88094286855773,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 42253
    },
    {
      "epoch": 16.88134238913304,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 42254
    },
    {
      "epoch": 16.88174190970835,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 42255
    },
    {
      "epoch": 16.88214143028366,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 42256
    },
    {
      "epoch": 16.88254095085897,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 42257
    },
    {
      "epoch": 16.882940471434278,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 42258
    },
    {
      "epoch": 16.88333999200959,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 42259
    },
    {
      "epoch": 16.8837395125849,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 42260
    },
    {
      "epoch": 16.884139033160206,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 42261
    },
    {
      "epoch": 16.884538553735517,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 42262
    },
    {
      "epoch": 16.884938074310828,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 42263
    },
    {
      "epoch": 16.88533759488614,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 42264
    },
    {
      "epoch": 16.885737115461446,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 42265
    },
    {
      "epoch": 16.886136636036756,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 42266
    },
    {
      "epoch": 16.886536156612067,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 42267
    },
    {
      "epoch": 16.886935677187374,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 42268
    },
    {
      "epoch": 16.887335197762685,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 42269
    },
    {
      "epoch": 16.887734718337995,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 42270
    },
    {
      "epoch": 16.888134238913302,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 42271
    },
    {
      "epoch": 16.888533759488613,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 42272
    },
    {
      "epoch": 16.888933280063924,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 42273
    },
    {
      "epoch": 16.889332800639234,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 42274
    },
    {
      "epoch": 16.88973232121454,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 42275
    },
    {
      "epoch": 16.890131841789852,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 42276
    },
    {
      "epoch": 16.890531362365163,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 42277
    },
    {
      "epoch": 16.89093088294047,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 42278
    },
    {
      "epoch": 16.89133040351578,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 42279
    },
    {
      "epoch": 16.89172992409109,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 42280
    },
    {
      "epoch": 16.892129444666402,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 42281
    },
    {
      "epoch": 16.89252896524171,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 42282
    },
    {
      "epoch": 16.89292848581702,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 42283
    },
    {
      "epoch": 16.89332800639233,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 42284
    },
    {
      "epoch": 16.893727526967638,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 42285
    },
    {
      "epoch": 16.89412704754295,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 42286
    },
    {
      "epoch": 16.89452656811826,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 42287
    },
    {
      "epoch": 16.894926088693566,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 42288
    },
    {
      "epoch": 16.895325609268877,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 42289
    },
    {
      "epoch": 16.895725129844188,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 42290
    },
    {
      "epoch": 16.896124650419498,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 42291
    },
    {
      "epoch": 16.896524170994805,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 42292
    },
    {
      "epoch": 16.896923691570116,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 42293
    },
    {
      "epoch": 16.897323212145427,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 42294
    },
    {
      "epoch": 16.897722732720734,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 42295
    },
    {
      "epoch": 16.898122253296044,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 42296
    },
    {
      "epoch": 16.898521773871355,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 42297
    },
    {
      "epoch": 16.898921294446666,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0124,
      "step": 42298
    },
    {
      "epoch": 16.899320815021973,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 42299
    },
    {
      "epoch": 16.899720335597284,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 42300
    },
    {
      "epoch": 16.900119856172594,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0072,
      "step": 42301
    },
    {
      "epoch": 16.9005193767479,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 42302
    },
    {
      "epoch": 16.900918897323212,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 42303
    },
    {
      "epoch": 16.901318417898523,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 42304
    },
    {
      "epoch": 16.90171793847383,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 42305
    },
    {
      "epoch": 16.90211745904914,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 42306
    },
    {
      "epoch": 16.90251697962445,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 42307
    },
    {
      "epoch": 16.902916500199762,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 42308
    },
    {
      "epoch": 16.90331602077507,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 42309
    },
    {
      "epoch": 16.90371554135038,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 42310
    },
    {
      "epoch": 16.90411506192569,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 42311
    },
    {
      "epoch": 16.904514582500997,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 42312
    },
    {
      "epoch": 16.904914103076308,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 42313
    },
    {
      "epoch": 16.90531362365162,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 42314
    },
    {
      "epoch": 16.90571314422693,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 42315
    },
    {
      "epoch": 16.906112664802237,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 42316
    },
    {
      "epoch": 16.906512185377547,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 42317
    },
    {
      "epoch": 16.906911705952858,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 42318
    },
    {
      "epoch": 16.907311226528165,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 42319
    },
    {
      "epoch": 16.907710747103476,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 42320
    },
    {
      "epoch": 16.908110267678786,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 42321
    },
    {
      "epoch": 16.908509788254094,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0445,
      "step": 42322
    },
    {
      "epoch": 16.908909308829404,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 42323
    },
    {
      "epoch": 16.909308829404715,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 42324
    },
    {
      "epoch": 16.909708349980026,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 42325
    },
    {
      "epoch": 16.910107870555333,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 42326
    },
    {
      "epoch": 16.910507391130643,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 42327
    },
    {
      "epoch": 16.910906911705954,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 42328
    },
    {
      "epoch": 16.91130643228126,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 42329
    },
    {
      "epoch": 16.911705952856572,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 42330
    },
    {
      "epoch": 16.912105473431883,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 42331
    },
    {
      "epoch": 16.91250499400719,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 42332
    },
    {
      "epoch": 16.9129045145825,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 42333
    },
    {
      "epoch": 16.91330403515781,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 42334
    },
    {
      "epoch": 16.91370355573312,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 42335
    },
    {
      "epoch": 16.91410307630843,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 42336
    },
    {
      "epoch": 16.91450259688374,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 42337
    },
    {
      "epoch": 16.91490211745905,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 42338
    },
    {
      "epoch": 16.915301638034357,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 42339
    },
    {
      "epoch": 16.915701158609668,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 42340
    },
    {
      "epoch": 16.91610067918498,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 42341
    },
    {
      "epoch": 16.91650019976029,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 42342
    },
    {
      "epoch": 16.916899720335596,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 42343
    },
    {
      "epoch": 16.917299240910907,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 42344
    },
    {
      "epoch": 16.917698761486218,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 42345
    },
    {
      "epoch": 16.918098282061525,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 42346
    },
    {
      "epoch": 16.918497802636836,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 42347
    },
    {
      "epoch": 16.918897323212146,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 42348
    },
    {
      "epoch": 16.919296843787453,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 42349
    },
    {
      "epoch": 16.919696364362764,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 42350
    },
    {
      "epoch": 16.920095884938075,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 42351
    },
    {
      "epoch": 16.920495405513385,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 42352
    },
    {
      "epoch": 16.920894926088693,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 42353
    },
    {
      "epoch": 16.921294446664003,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 42354
    },
    {
      "epoch": 16.921693967239314,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 42355
    },
    {
      "epoch": 16.92209348781462,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 42356
    },
    {
      "epoch": 16.92249300838993,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 42357
    },
    {
      "epoch": 16.922892528965242,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 42358
    },
    {
      "epoch": 16.923292049540553,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 42359
    },
    {
      "epoch": 16.92369157011586,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 42360
    },
    {
      "epoch": 16.92409109069117,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 42361
    },
    {
      "epoch": 16.92449061126648,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 42362
    },
    {
      "epoch": 16.92489013184179,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 42363
    },
    {
      "epoch": 16.9252896524171,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 42364
    },
    {
      "epoch": 16.92568917299241,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 42365
    },
    {
      "epoch": 16.926088693567717,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 42366
    },
    {
      "epoch": 16.926488214143028,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 42367
    },
    {
      "epoch": 16.92688773471834,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 42368
    },
    {
      "epoch": 16.92728725529365,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 42369
    },
    {
      "epoch": 16.927686775868956,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 42370
    },
    {
      "epoch": 16.928086296444267,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 42371
    },
    {
      "epoch": 16.928485817019578,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 42372
    },
    {
      "epoch": 16.928885337594885,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 42373
    },
    {
      "epoch": 16.929284858170195,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 42374
    },
    {
      "epoch": 16.929684378745506,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 42375
    },
    {
      "epoch": 16.930083899320817,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 42376
    },
    {
      "epoch": 16.930483419896124,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 42377
    },
    {
      "epoch": 16.930882940471434,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 42378
    },
    {
      "epoch": 16.931282461046745,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 42379
    },
    {
      "epoch": 16.931681981622052,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 42380
    },
    {
      "epoch": 16.932081502197363,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 42381
    },
    {
      "epoch": 16.932481022772674,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 42382
    },
    {
      "epoch": 16.93288054334798,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 42383
    },
    {
      "epoch": 16.93328006392329,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 42384
    },
    {
      "epoch": 16.933679584498602,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 42385
    },
    {
      "epoch": 16.934079105073913,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0464,
      "step": 42386
    },
    {
      "epoch": 16.93447862564922,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 42387
    },
    {
      "epoch": 16.93487814622453,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 42388
    },
    {
      "epoch": 16.93527766679984,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 42389
    },
    {
      "epoch": 16.93567718737515,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 42390
    },
    {
      "epoch": 16.93607670795046,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 42391
    },
    {
      "epoch": 16.93647622852577,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 42392
    },
    {
      "epoch": 16.93687574910108,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 42393
    },
    {
      "epoch": 16.937275269676388,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 42394
    },
    {
      "epoch": 16.937674790251698,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 42395
    },
    {
      "epoch": 16.93807431082701,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 42396
    },
    {
      "epoch": 16.938473831402316,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0135,
      "step": 42397
    },
    {
      "epoch": 16.938873351977627,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 42398
    },
    {
      "epoch": 16.939272872552937,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 42399
    },
    {
      "epoch": 16.939672393128244,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 42400
    },
    {
      "epoch": 16.940071913703555,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 42401
    },
    {
      "epoch": 16.940471434278866,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 42402
    },
    {
      "epoch": 16.940870954854176,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 42403
    },
    {
      "epoch": 16.941270475429484,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 42404
    },
    {
      "epoch": 16.941669996004794,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 42405
    },
    {
      "epoch": 16.942069516580105,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 42406
    },
    {
      "epoch": 16.942469037155412,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 42407
    },
    {
      "epoch": 16.942868557730723,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 42408
    },
    {
      "epoch": 16.943268078306033,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 42409
    },
    {
      "epoch": 16.943667598881344,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 42410
    },
    {
      "epoch": 16.94406711945665,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 42411
    },
    {
      "epoch": 16.944466640031962,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 42412
    },
    {
      "epoch": 16.944866160607273,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 42413
    },
    {
      "epoch": 16.94526568118258,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 42414
    },
    {
      "epoch": 16.94566520175789,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 42415
    },
    {
      "epoch": 16.9460647223332,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 42416
    },
    {
      "epoch": 16.946464242908508,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 42417
    },
    {
      "epoch": 16.94686376348382,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 42418
    },
    {
      "epoch": 16.94726328405913,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 42419
    },
    {
      "epoch": 16.94766280463444,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 42420
    },
    {
      "epoch": 16.948062325209747,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 42421
    },
    {
      "epoch": 16.948461845785058,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 42422
    },
    {
      "epoch": 16.94886136636037,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 42423
    },
    {
      "epoch": 16.949260886935676,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 42424
    },
    {
      "epoch": 16.949660407510986,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 42425
    },
    {
      "epoch": 16.950059928086297,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 42426
    },
    {
      "epoch": 16.950459448661604,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0102,
      "step": 42427
    },
    {
      "epoch": 16.950858969236915,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 42428
    },
    {
      "epoch": 16.951258489812226,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 42429
    },
    {
      "epoch": 16.951658010387536,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 42430
    },
    {
      "epoch": 16.952057530962843,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 42431
    },
    {
      "epoch": 16.952457051538154,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 42432
    },
    {
      "epoch": 16.952856572113465,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 42433
    },
    {
      "epoch": 16.953256092688772,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 42434
    },
    {
      "epoch": 16.953655613264083,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 42435
    },
    {
      "epoch": 16.954055133839393,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 42436
    },
    {
      "epoch": 16.954454654414704,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 42437
    },
    {
      "epoch": 16.95485417499001,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 42438
    },
    {
      "epoch": 16.95525369556532,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 42439
    },
    {
      "epoch": 16.955653216140632,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 42440
    },
    {
      "epoch": 16.95605273671594,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 42441
    },
    {
      "epoch": 16.95645225729125,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 42442
    },
    {
      "epoch": 16.95685177786656,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 42443
    },
    {
      "epoch": 16.957251298441868,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 42444
    },
    {
      "epoch": 16.95765081901718,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 42445
    },
    {
      "epoch": 16.95805033959249,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 42446
    },
    {
      "epoch": 16.9584498601678,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 42447
    },
    {
      "epoch": 16.958849380743107,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 42448
    },
    {
      "epoch": 16.959248901318418,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0087,
      "step": 42449
    },
    {
      "epoch": 16.95964842189373,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 42450
    },
    {
      "epoch": 16.960047942469036,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 42451
    },
    {
      "epoch": 16.960447463044346,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 42452
    },
    {
      "epoch": 16.960846983619657,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 42453
    },
    {
      "epoch": 16.961246504194968,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 42454
    },
    {
      "epoch": 16.961646024770275,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 42455
    },
    {
      "epoch": 16.962045545345585,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 42456
    },
    {
      "epoch": 16.962445065920896,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 42457
    },
    {
      "epoch": 16.962844586496203,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 42458
    },
    {
      "epoch": 16.963244107071514,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 42459
    },
    {
      "epoch": 16.963643627646825,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 42460
    },
    {
      "epoch": 16.96404314822213,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 42461
    },
    {
      "epoch": 16.964442668797442,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 42462
    },
    {
      "epoch": 16.964842189372753,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 42463
    },
    {
      "epoch": 16.965241709948064,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 42464
    },
    {
      "epoch": 16.96564123052337,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 42465
    },
    {
      "epoch": 16.96604075109868,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 42466
    },
    {
      "epoch": 16.966440271673992,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 42467
    },
    {
      "epoch": 16.9668397922493,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 42468
    },
    {
      "epoch": 16.96723931282461,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 42469
    },
    {
      "epoch": 16.96763883339992,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 42470
    },
    {
      "epoch": 16.96803835397523,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 42471
    },
    {
      "epoch": 16.96843787455054,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 42472
    },
    {
      "epoch": 16.96883739512585,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 42473
    },
    {
      "epoch": 16.96923691570116,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 42474
    },
    {
      "epoch": 16.969636436276467,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 42475
    },
    {
      "epoch": 16.970035956851778,
      "grad_norm": 0.265625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 42476
    },
    {
      "epoch": 16.97043547742709,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 42477
    },
    {
      "epoch": 16.970834998002395,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 42478
    },
    {
      "epoch": 16.971234518577706,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 42479
    },
    {
      "epoch": 16.971634039153017,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 42480
    },
    {
      "epoch": 16.972033559728327,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 42481
    },
    {
      "epoch": 16.972433080303635,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 42482
    },
    {
      "epoch": 16.972832600878945,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0077,
      "step": 42483
    },
    {
      "epoch": 16.973232121454256,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 42484
    },
    {
      "epoch": 16.973631642029563,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 42485
    },
    {
      "epoch": 16.974031162604874,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 42486
    },
    {
      "epoch": 16.974430683180184,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 42487
    },
    {
      "epoch": 16.974830203755495,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 42488
    },
    {
      "epoch": 16.975229724330802,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 42489
    },
    {
      "epoch": 16.975629244906113,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 42490
    },
    {
      "epoch": 16.976028765481423,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 42491
    },
    {
      "epoch": 16.97642828605673,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 42492
    },
    {
      "epoch": 16.97682780663204,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 42493
    },
    {
      "epoch": 16.977227327207352,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 42494
    },
    {
      "epoch": 16.97762684778266,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 42495
    },
    {
      "epoch": 16.97802636835797,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 42496
    },
    {
      "epoch": 16.97842588893328,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 42497
    },
    {
      "epoch": 16.97882540950859,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 42498
    },
    {
      "epoch": 16.979224930083898,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 42499
    },
    {
      "epoch": 16.97962445065921,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 42500
    },
    {
      "epoch": 16.98002397123452,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 42501
    },
    {
      "epoch": 16.980423491809827,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 42502
    },
    {
      "epoch": 16.980823012385137,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 42503
    },
    {
      "epoch": 16.981222532960448,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 42504
    },
    {
      "epoch": 16.98162205353576,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 42505
    },
    {
      "epoch": 16.982021574111066,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 42506
    },
    {
      "epoch": 16.982421094686377,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 42507
    },
    {
      "epoch": 16.982820615261687,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0124,
      "step": 42508
    },
    {
      "epoch": 16.983220135836994,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 42509
    },
    {
      "epoch": 16.983619656412305,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 42510
    },
    {
      "epoch": 16.984019176987616,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 42511
    },
    {
      "epoch": 16.984418697562923,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 42512
    },
    {
      "epoch": 16.984818218138233,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 42513
    },
    {
      "epoch": 16.985217738713544,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 42514
    },
    {
      "epoch": 16.985617259288855,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 42515
    },
    {
      "epoch": 16.986016779864162,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 42516
    },
    {
      "epoch": 16.986416300439473,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 42517
    },
    {
      "epoch": 16.986815821014783,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 42518
    },
    {
      "epoch": 16.98721534159009,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 42519
    },
    {
      "epoch": 16.9876148621654,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 42520
    },
    {
      "epoch": 16.98801438274071,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0115,
      "step": 42521
    },
    {
      "epoch": 16.988413903316022,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 42522
    },
    {
      "epoch": 16.98881342389133,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 42523
    },
    {
      "epoch": 16.98921294446664,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 42524
    },
    {
      "epoch": 16.98961246504195,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 42525
    },
    {
      "epoch": 16.990011985617258,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 42526
    },
    {
      "epoch": 16.99041150619257,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 42527
    },
    {
      "epoch": 16.99081102676788,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 42528
    },
    {
      "epoch": 16.991210547343186,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 42529
    },
    {
      "epoch": 16.991610067918497,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 42530
    },
    {
      "epoch": 16.992009588493808,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 42531
    },
    {
      "epoch": 16.99240910906912,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 42532
    },
    {
      "epoch": 16.992808629644426,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 42533
    },
    {
      "epoch": 16.993208150219736,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 42534
    },
    {
      "epoch": 16.993607670795047,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 42535
    },
    {
      "epoch": 16.994007191370354,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 42536
    },
    {
      "epoch": 16.994406711945665,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 42537
    },
    {
      "epoch": 16.994806232520975,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 42538
    },
    {
      "epoch": 16.995205753096286,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 42539
    },
    {
      "epoch": 16.995605273671593,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 42540
    },
    {
      "epoch": 16.996004794246904,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 42541
    },
    {
      "epoch": 16.996404314822215,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 42542
    },
    {
      "epoch": 16.99680383539752,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 42543
    },
    {
      "epoch": 16.997203355972832,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 42544
    },
    {
      "epoch": 16.997602876548143,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 42545
    },
    {
      "epoch": 16.99800239712345,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 42546
    },
    {
      "epoch": 16.99840191769876,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 42547
    },
    {
      "epoch": 16.99880143827407,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 42548
    },
    {
      "epoch": 16.999200958849382,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 42549
    },
    {
      "epoch": 16.99960047942469,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 42550
    },
    {
      "epoch": 17.0,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 42551
    },
    {
      "epoch": 17.00039952057531,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 42552
    },
    {
      "epoch": 17.000799041150618,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 42553
    },
    {
      "epoch": 17.00119856172593,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 42554
    },
    {
      "epoch": 17.00159808230124,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 42555
    },
    {
      "epoch": 17.00199760287655,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 42556
    },
    {
      "epoch": 17.002397123451857,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 42557
    },
    {
      "epoch": 17.002796644027168,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 42558
    },
    {
      "epoch": 17.00319616460248,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 42559
    },
    {
      "epoch": 17.003595685177785,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 42560
    },
    {
      "epoch": 17.003995205753096,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 42561
    },
    {
      "epoch": 17.004394726328407,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 42562
    },
    {
      "epoch": 17.004794246903714,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 42563
    },
    {
      "epoch": 17.005193767479025,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 42564
    },
    {
      "epoch": 17.005593288054335,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 42565
    },
    {
      "epoch": 17.005992808629646,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 42566
    },
    {
      "epoch": 17.006392329204953,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 42567
    },
    {
      "epoch": 17.006791849780264,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 42568
    },
    {
      "epoch": 17.007191370355574,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 42569
    },
    {
      "epoch": 17.00759089093088,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 42570
    },
    {
      "epoch": 17.007990411506192,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 42571
    },
    {
      "epoch": 17.008389932081503,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 42572
    },
    {
      "epoch": 17.008789452656814,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 42573
    },
    {
      "epoch": 17.00918897323212,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 42574
    },
    {
      "epoch": 17.00958849380743,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 42575
    },
    {
      "epoch": 17.009988014382742,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 42576
    },
    {
      "epoch": 17.01038753495805,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 42577
    },
    {
      "epoch": 17.01078705553336,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 42578
    },
    {
      "epoch": 17.01118657610867,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 42579
    },
    {
      "epoch": 17.011586096683978,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 42580
    },
    {
      "epoch": 17.01198561725929,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 42581
    },
    {
      "epoch": 17.0123851378346,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 42582
    },
    {
      "epoch": 17.01278465840991,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 42583
    },
    {
      "epoch": 17.013184178985217,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 42584
    },
    {
      "epoch": 17.013583699560527,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 42585
    },
    {
      "epoch": 17.013983220135838,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 42586
    },
    {
      "epoch": 17.014382740711145,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 42587
    },
    {
      "epoch": 17.014782261286456,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 42588
    },
    {
      "epoch": 17.015181781861767,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 42589
    },
    {
      "epoch": 17.015581302437077,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 42590
    },
    {
      "epoch": 17.015980823012384,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 42591
    },
    {
      "epoch": 17.016380343587695,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 42592
    },
    {
      "epoch": 17.016779864163006,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 42593
    },
    {
      "epoch": 17.017179384738313,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 42594
    },
    {
      "epoch": 17.017578905313623,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 42595
    },
    {
      "epoch": 17.017978425888934,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 42596
    },
    {
      "epoch": 17.01837794646424,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 42597
    },
    {
      "epoch": 17.018777467039552,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 42598
    },
    {
      "epoch": 17.019176987614863,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 42599
    },
    {
      "epoch": 17.019576508190173,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 42600
    },
    {
      "epoch": 17.01997602876548,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 42601
    },
    {
      "epoch": 17.02037554934079,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 42602
    },
    {
      "epoch": 17.020775069916102,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 42603
    },
    {
      "epoch": 17.02117459049141,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 42604
    },
    {
      "epoch": 17.02157411106672,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 42605
    },
    {
      "epoch": 17.02197363164203,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 42606
    },
    {
      "epoch": 17.02237315221734,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 42607
    },
    {
      "epoch": 17.022772672792648,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 42608
    },
    {
      "epoch": 17.02317219336796,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 42609
    },
    {
      "epoch": 17.02357171394327,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 42610
    },
    {
      "epoch": 17.023971234518577,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 42611
    },
    {
      "epoch": 17.024370755093887,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 42612
    },
    {
      "epoch": 17.024770275669198,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 42613
    },
    {
      "epoch": 17.025169796244505,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 42614
    },
    {
      "epoch": 17.025569316819816,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 42615
    },
    {
      "epoch": 17.025968837395126,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 42616
    },
    {
      "epoch": 17.026368357970437,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 42617
    },
    {
      "epoch": 17.026767878545744,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 42618
    },
    {
      "epoch": 17.027167399121055,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 42619
    },
    {
      "epoch": 17.027566919696365,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 42620
    },
    {
      "epoch": 17.027966440271673,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 42621
    },
    {
      "epoch": 17.028365960846983,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 42622
    },
    {
      "epoch": 17.028765481422294,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 42623
    },
    {
      "epoch": 17.029165001997605,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 42624
    },
    {
      "epoch": 17.02956452257291,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 42625
    },
    {
      "epoch": 17.029964043148222,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 42626
    },
    {
      "epoch": 17.030363563723533,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 42627
    },
    {
      "epoch": 17.03076308429884,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 42628
    },
    {
      "epoch": 17.03116260487415,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 42629
    },
    {
      "epoch": 17.03156212544946,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 42630
    },
    {
      "epoch": 17.03196164602477,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 42631
    },
    {
      "epoch": 17.03236116660008,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 42632
    },
    {
      "epoch": 17.03276068717539,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 42633
    },
    {
      "epoch": 17.0331602077507,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 42634
    },
    {
      "epoch": 17.033559728326008,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 42635
    },
    {
      "epoch": 17.03395924890132,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 42636
    },
    {
      "epoch": 17.03435876947663,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 42637
    },
    {
      "epoch": 17.034758290051936,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 42638
    },
    {
      "epoch": 17.035157810627247,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 42639
    },
    {
      "epoch": 17.035557331202558,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 42640
    },
    {
      "epoch": 17.03595685177787,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 42641
    },
    {
      "epoch": 17.036356372353175,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 42642
    },
    {
      "epoch": 17.036755892928486,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 42643
    },
    {
      "epoch": 17.037155413503797,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 42644
    },
    {
      "epoch": 17.037554934079104,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 42645
    },
    {
      "epoch": 17.037954454654415,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 42646
    },
    {
      "epoch": 17.038353975229725,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 42647
    },
    {
      "epoch": 17.038753495805032,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 42648
    },
    {
      "epoch": 17.039153016380343,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 42649
    },
    {
      "epoch": 17.039552536955654,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 42650
    },
    {
      "epoch": 17.039952057530964,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 42651
    },
    {
      "epoch": 17.04035157810627,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 42652
    },
    {
      "epoch": 17.040751098681582,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 42653
    },
    {
      "epoch": 17.041150619256893,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 42654
    },
    {
      "epoch": 17.0415501398322,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 42655
    },
    {
      "epoch": 17.04194966040751,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 42656
    },
    {
      "epoch": 17.04234918098282,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 42657
    },
    {
      "epoch": 17.04274870155813,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 42658
    },
    {
      "epoch": 17.04314822213344,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 42659
    },
    {
      "epoch": 17.04354774270875,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 42660
    },
    {
      "epoch": 17.04394726328406,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 42661
    },
    {
      "epoch": 17.044346783859368,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 42662
    },
    {
      "epoch": 17.04474630443468,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 42663
    },
    {
      "epoch": 17.04514582500999,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 42664
    },
    {
      "epoch": 17.045545345585296,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 42665
    },
    {
      "epoch": 17.045944866160607,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 42666
    },
    {
      "epoch": 17.046344386735917,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 42667
    },
    {
      "epoch": 17.046743907311228,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 42668
    },
    {
      "epoch": 17.047143427886535,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 42669
    },
    {
      "epoch": 17.047542948461846,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 42670
    },
    {
      "epoch": 17.047942469037157,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 42671
    },
    {
      "epoch": 17.048341989612464,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 42672
    },
    {
      "epoch": 17.048741510187774,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 42673
    },
    {
      "epoch": 17.049141030763085,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 42674
    },
    {
      "epoch": 17.049540551338392,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 42675
    },
    {
      "epoch": 17.049940071913703,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 42676
    },
    {
      "epoch": 17.050339592489014,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 42677
    },
    {
      "epoch": 17.050739113064324,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 42678
    },
    {
      "epoch": 17.05113863363963,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 42679
    },
    {
      "epoch": 17.051538154214942,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 42680
    },
    {
      "epoch": 17.051937674790253,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 42681
    },
    {
      "epoch": 17.05233719536556,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0405,
      "step": 42682
    },
    {
      "epoch": 17.05273671594087,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 42683
    },
    {
      "epoch": 17.05313623651618,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 42684
    },
    {
      "epoch": 17.053535757091492,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 42685
    },
    {
      "epoch": 17.0539352776668,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 42686
    },
    {
      "epoch": 17.05433479824211,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 42687
    },
    {
      "epoch": 17.05473431881742,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 42688
    },
    {
      "epoch": 17.055133839392727,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 42689
    },
    {
      "epoch": 17.055533359968038,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 42690
    },
    {
      "epoch": 17.05593288054335,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 42691
    },
    {
      "epoch": 17.056332401118656,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 42692
    },
    {
      "epoch": 17.056731921693967,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 42693
    },
    {
      "epoch": 17.057131442269277,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 42694
    },
    {
      "epoch": 17.057530962844588,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 42695
    },
    {
      "epoch": 17.057930483419895,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 42696
    },
    {
      "epoch": 17.058330003995206,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 42697
    },
    {
      "epoch": 17.058729524570516,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 42698
    },
    {
      "epoch": 17.059129045145824,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 42699
    },
    {
      "epoch": 17.059528565721134,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 42700
    },
    {
      "epoch": 17.059928086296445,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 42701
    },
    {
      "epoch": 17.060327606871756,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 42702
    },
    {
      "epoch": 17.060727127447063,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 42703
    },
    {
      "epoch": 17.061126648022373,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 42704
    },
    {
      "epoch": 17.061526168597684,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 42705
    },
    {
      "epoch": 17.06192568917299,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 42706
    },
    {
      "epoch": 17.062325209748302,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 42707
    },
    {
      "epoch": 17.062724730323612,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 42708
    },
    {
      "epoch": 17.06312425089892,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 42709
    },
    {
      "epoch": 17.06352377147423,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 42710
    },
    {
      "epoch": 17.06392329204954,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 42711
    },
    {
      "epoch": 17.06432281262485,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0131,
      "step": 42712
    },
    {
      "epoch": 17.06472233320016,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 42713
    },
    {
      "epoch": 17.06512185377547,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 42714
    },
    {
      "epoch": 17.06552137435078,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 42715
    },
    {
      "epoch": 17.065920894926087,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 42716
    },
    {
      "epoch": 17.066320415501398,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 42717
    },
    {
      "epoch": 17.06671993607671,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 42718
    },
    {
      "epoch": 17.06711945665202,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0128,
      "step": 42719
    },
    {
      "epoch": 17.067518977227326,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 42720
    },
    {
      "epoch": 17.067918497802637,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 42721
    },
    {
      "epoch": 17.068318018377948,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 42722
    },
    {
      "epoch": 17.068717538953255,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 42723
    },
    {
      "epoch": 17.069117059528566,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 42724
    },
    {
      "epoch": 17.069516580103876,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 42725
    },
    {
      "epoch": 17.069916100679183,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 42726
    },
    {
      "epoch": 17.070315621254494,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 42727
    },
    {
      "epoch": 17.070715141829805,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 42728
    },
    {
      "epoch": 17.071114662405115,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 42729
    },
    {
      "epoch": 17.071514182980422,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 42730
    },
    {
      "epoch": 17.071913703555733,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 42731
    },
    {
      "epoch": 17.072313224131044,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 42732
    },
    {
      "epoch": 17.07271274470635,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 42733
    },
    {
      "epoch": 17.07311226528166,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 42734
    },
    {
      "epoch": 17.073511785856972,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 42735
    },
    {
      "epoch": 17.073911306432283,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 42736
    },
    {
      "epoch": 17.07431082700759,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 42737
    },
    {
      "epoch": 17.0747103475829,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 42738
    },
    {
      "epoch": 17.07510986815821,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 42739
    },
    {
      "epoch": 17.07550938873352,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 42740
    },
    {
      "epoch": 17.07590890930883,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 42741
    },
    {
      "epoch": 17.07630842988414,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 42742
    },
    {
      "epoch": 17.076707950459447,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 42743
    },
    {
      "epoch": 17.077107471034758,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 42744
    },
    {
      "epoch": 17.07750699161007,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 42745
    },
    {
      "epoch": 17.07790651218538,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 42746
    },
    {
      "epoch": 17.078306032760686,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 42747
    },
    {
      "epoch": 17.078705553335997,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 42748
    },
    {
      "epoch": 17.079105073911307,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 42749
    },
    {
      "epoch": 17.079504594486615,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 42750
    },
    {
      "epoch": 17.079904115061925,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 42751
    },
    {
      "epoch": 17.080303635637236,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0471,
      "step": 42752
    },
    {
      "epoch": 17.080703156212547,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 42753
    },
    {
      "epoch": 17.081102676787854,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 42754
    },
    {
      "epoch": 17.081502197363164,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 42755
    },
    {
      "epoch": 17.081901717938475,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 42756
    },
    {
      "epoch": 17.082301238513782,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 42757
    },
    {
      "epoch": 17.082700759089093,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 42758
    },
    {
      "epoch": 17.083100279664404,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 42759
    },
    {
      "epoch": 17.08349980023971,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 42760
    },
    {
      "epoch": 17.08389932081502,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 42761
    },
    {
      "epoch": 17.084298841390332,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 42762
    },
    {
      "epoch": 17.084698361965643,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 42763
    },
    {
      "epoch": 17.08509788254095,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 42764
    },
    {
      "epoch": 17.08549740311626,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 42765
    },
    {
      "epoch": 17.08589692369157,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0125,
      "step": 42766
    },
    {
      "epoch": 17.08629644426688,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 42767
    },
    {
      "epoch": 17.08669596484219,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 42768
    },
    {
      "epoch": 17.0870954854175,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 42769
    },
    {
      "epoch": 17.08749500599281,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 42770
    },
    {
      "epoch": 17.087894526568117,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 42771
    },
    {
      "epoch": 17.088294047143428,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 42772
    },
    {
      "epoch": 17.08869356771874,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 42773
    },
    {
      "epoch": 17.089093088294046,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 42774
    },
    {
      "epoch": 17.089492608869357,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 42775
    },
    {
      "epoch": 17.089892129444667,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 42776
    },
    {
      "epoch": 17.090291650019974,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 42777
    },
    {
      "epoch": 17.090691170595285,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 42778
    },
    {
      "epoch": 17.091090691170596,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 42779
    },
    {
      "epoch": 17.091490211745906,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 42780
    },
    {
      "epoch": 17.091889732321214,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 42781
    },
    {
      "epoch": 17.092289252896524,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 42782
    },
    {
      "epoch": 17.092688773471835,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 42783
    },
    {
      "epoch": 17.093088294047142,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 42784
    },
    {
      "epoch": 17.093487814622453,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 42785
    },
    {
      "epoch": 17.093887335197763,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 42786
    },
    {
      "epoch": 17.094286855773074,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 42787
    },
    {
      "epoch": 17.09468637634838,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 42788
    },
    {
      "epoch": 17.095085896923692,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.045,
      "step": 42789
    },
    {
      "epoch": 17.095485417499003,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 42790
    },
    {
      "epoch": 17.09588493807431,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 42791
    },
    {
      "epoch": 17.09628445864962,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 42792
    },
    {
      "epoch": 17.09668397922493,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 42793
    },
    {
      "epoch": 17.097083499800238,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 42794
    },
    {
      "epoch": 17.09748302037555,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 42795
    },
    {
      "epoch": 17.09788254095086,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 42796
    },
    {
      "epoch": 17.09828206152617,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 42797
    },
    {
      "epoch": 17.098681582101477,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 42798
    },
    {
      "epoch": 17.099081102676788,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 42799
    },
    {
      "epoch": 17.0994806232521,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0106,
      "step": 42800
    },
    {
      "epoch": 17.099880143827406,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 42801
    },
    {
      "epoch": 17.100279664402716,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 42802
    },
    {
      "epoch": 17.100679184978027,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 42803
    },
    {
      "epoch": 17.101078705553338,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 42804
    },
    {
      "epoch": 17.101478226128645,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 42805
    },
    {
      "epoch": 17.101877746703956,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 42806
    },
    {
      "epoch": 17.102277267279266,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 42807
    },
    {
      "epoch": 17.102676787854573,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 42808
    },
    {
      "epoch": 17.103076308429884,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 42809
    },
    {
      "epoch": 17.103475829005195,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 42810
    },
    {
      "epoch": 17.103875349580502,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 42811
    },
    {
      "epoch": 17.104274870155812,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 42812
    },
    {
      "epoch": 17.104674390731123,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 42813
    },
    {
      "epoch": 17.105073911306434,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 42814
    },
    {
      "epoch": 17.10547343188174,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 42815
    },
    {
      "epoch": 17.10587295245705,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 42816
    },
    {
      "epoch": 17.106272473032362,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 42817
    },
    {
      "epoch": 17.10667199360767,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 42818
    },
    {
      "epoch": 17.10707151418298,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 42819
    },
    {
      "epoch": 17.10747103475829,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 42820
    },
    {
      "epoch": 17.107870555333598,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 42821
    },
    {
      "epoch": 17.10827007590891,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 42822
    },
    {
      "epoch": 17.10866959648422,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 42823
    },
    {
      "epoch": 17.10906911705953,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 42824
    },
    {
      "epoch": 17.109468637634837,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 42825
    },
    {
      "epoch": 17.109868158210148,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 42826
    },
    {
      "epoch": 17.11026767878546,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 42827
    },
    {
      "epoch": 17.110667199360766,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 42828
    },
    {
      "epoch": 17.111066719936076,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 42829
    },
    {
      "epoch": 17.111466240511387,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 42830
    },
    {
      "epoch": 17.111865761086698,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 42831
    },
    {
      "epoch": 17.112265281662005,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 42832
    },
    {
      "epoch": 17.112664802237315,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 42833
    },
    {
      "epoch": 17.113064322812626,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 42834
    },
    {
      "epoch": 17.113463843387933,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 42835
    },
    {
      "epoch": 17.113863363963244,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 42836
    },
    {
      "epoch": 17.114262884538554,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 42837
    },
    {
      "epoch": 17.11466240511386,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 42838
    },
    {
      "epoch": 17.115061925689172,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 42839
    },
    {
      "epoch": 17.115461446264483,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 42840
    },
    {
      "epoch": 17.115860966839794,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 42841
    },
    {
      "epoch": 17.1162604874151,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 42842
    },
    {
      "epoch": 17.11666000799041,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 42843
    },
    {
      "epoch": 17.117059528565722,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 42844
    },
    {
      "epoch": 17.11745904914103,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 42845
    },
    {
      "epoch": 17.11785856971634,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 42846
    },
    {
      "epoch": 17.11825809029165,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 42847
    },
    {
      "epoch": 17.11865761086696,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 42848
    },
    {
      "epoch": 17.11905713144227,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 42849
    },
    {
      "epoch": 17.11945665201758,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0126,
      "step": 42850
    },
    {
      "epoch": 17.11985617259289,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 42851
    },
    {
      "epoch": 17.120255693168197,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 42852
    },
    {
      "epoch": 17.120655213743508,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 42853
    },
    {
      "epoch": 17.121054734318818,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 42854
    },
    {
      "epoch": 17.121454254894125,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 42855
    },
    {
      "epoch": 17.121853775469436,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 42856
    },
    {
      "epoch": 17.122253296044747,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 42857
    },
    {
      "epoch": 17.122652816620057,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 42858
    },
    {
      "epoch": 17.123052337195364,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 42859
    },
    {
      "epoch": 17.123451857770675,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 42860
    },
    {
      "epoch": 17.123851378345986,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 42861
    },
    {
      "epoch": 17.124250898921293,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 42862
    },
    {
      "epoch": 17.124650419496604,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 42863
    },
    {
      "epoch": 17.125049940071914,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 42864
    },
    {
      "epoch": 17.125449460647225,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 42865
    },
    {
      "epoch": 17.125848981222532,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 42866
    },
    {
      "epoch": 17.126248501797843,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 42867
    },
    {
      "epoch": 17.126648022373153,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 42868
    },
    {
      "epoch": 17.12704754294846,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 42869
    },
    {
      "epoch": 17.12744706352377,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 42870
    },
    {
      "epoch": 17.127846584099082,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 42871
    },
    {
      "epoch": 17.12824610467439,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0119,
      "step": 42872
    },
    {
      "epoch": 17.1286456252497,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 42873
    },
    {
      "epoch": 17.12904514582501,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 42874
    },
    {
      "epoch": 17.12944466640032,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 42875
    },
    {
      "epoch": 17.129844186975628,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 42876
    },
    {
      "epoch": 17.13024370755094,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 42877
    },
    {
      "epoch": 17.13064322812625,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 42878
    },
    {
      "epoch": 17.131042748701557,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 42879
    },
    {
      "epoch": 17.131442269276867,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 42880
    },
    {
      "epoch": 17.131841789852178,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 42881
    },
    {
      "epoch": 17.13224131042749,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 42882
    },
    {
      "epoch": 17.132640831002796,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0134,
      "step": 42883
    },
    {
      "epoch": 17.133040351578106,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 42884
    },
    {
      "epoch": 17.133439872153417,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 42885
    },
    {
      "epoch": 17.133839392728724,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 42886
    },
    {
      "epoch": 17.134238913304035,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 42887
    },
    {
      "epoch": 17.134638433879346,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 42888
    },
    {
      "epoch": 17.135037954454653,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 42889
    },
    {
      "epoch": 17.135437475029963,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 42890
    },
    {
      "epoch": 17.135836995605274,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 42891
    },
    {
      "epoch": 17.136236516180585,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 42892
    },
    {
      "epoch": 17.136636036755892,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 42893
    },
    {
      "epoch": 17.137035557331203,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 42894
    },
    {
      "epoch": 17.137435077906513,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 42895
    },
    {
      "epoch": 17.13783459848182,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 42896
    },
    {
      "epoch": 17.13823411905713,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 42897
    },
    {
      "epoch": 17.13863363963244,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 42898
    },
    {
      "epoch": 17.139033160207752,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 42899
    },
    {
      "epoch": 17.13943268078306,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 42900
    },
    {
      "epoch": 17.13983220135837,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 42901
    },
    {
      "epoch": 17.14023172193368,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 42902
    },
    {
      "epoch": 17.140631242508988,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 42903
    },
    {
      "epoch": 17.1410307630843,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 42904
    },
    {
      "epoch": 17.14143028365961,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0078,
      "step": 42905
    },
    {
      "epoch": 17.141829804234916,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 42906
    },
    {
      "epoch": 17.142229324810227,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 42907
    },
    {
      "epoch": 17.142628845385538,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 42908
    },
    {
      "epoch": 17.14302836596085,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 42909
    },
    {
      "epoch": 17.143427886536156,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 42910
    },
    {
      "epoch": 17.143827407111466,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 42911
    },
    {
      "epoch": 17.144226927686777,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 42912
    },
    {
      "epoch": 17.144626448262084,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 42913
    },
    {
      "epoch": 17.145025968837395,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 42914
    },
    {
      "epoch": 17.145425489412705,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 42915
    },
    {
      "epoch": 17.145825009988016,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 42916
    },
    {
      "epoch": 17.146224530563323,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 42917
    },
    {
      "epoch": 17.146624051138634,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 42918
    },
    {
      "epoch": 17.147023571713945,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 42919
    },
    {
      "epoch": 17.14742309228925,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 42920
    },
    {
      "epoch": 17.147822612864562,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 42921
    },
    {
      "epoch": 17.148222133439873,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 42922
    },
    {
      "epoch": 17.14862165401518,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 42923
    },
    {
      "epoch": 17.14902117459049,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 42924
    },
    {
      "epoch": 17.1494206951658,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 42925
    },
    {
      "epoch": 17.149820215741112,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 42926
    },
    {
      "epoch": 17.15021973631642,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 42927
    },
    {
      "epoch": 17.15061925689173,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 42928
    },
    {
      "epoch": 17.15101877746704,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 42929
    },
    {
      "epoch": 17.151418298042348,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 42930
    },
    {
      "epoch": 17.15181781861766,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 42931
    },
    {
      "epoch": 17.15221733919297,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 42932
    },
    {
      "epoch": 17.15261685976828,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 42933
    },
    {
      "epoch": 17.153016380343587,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0096,
      "step": 42934
    },
    {
      "epoch": 17.153415900918898,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 42935
    },
    {
      "epoch": 17.15381542149421,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 42936
    },
    {
      "epoch": 17.154214942069515,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 42937
    },
    {
      "epoch": 17.154614462644826,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 42938
    },
    {
      "epoch": 17.155013983220137,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 42939
    },
    {
      "epoch": 17.155413503795444,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 42940
    },
    {
      "epoch": 17.155813024370755,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 42941
    },
    {
      "epoch": 17.156212544946065,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 42942
    },
    {
      "epoch": 17.156612065521376,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 42943
    },
    {
      "epoch": 17.157011586096683,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0107,
      "step": 42944
    },
    {
      "epoch": 17.157411106671994,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0414,
      "step": 42945
    },
    {
      "epoch": 17.157810627247304,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 42946
    },
    {
      "epoch": 17.15821014782261,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 42947
    },
    {
      "epoch": 17.158609668397922,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 42948
    },
    {
      "epoch": 17.159009188973233,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0422,
      "step": 42949
    },
    {
      "epoch": 17.159408709548543,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 42950
    },
    {
      "epoch": 17.15980823012385,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 42951
    },
    {
      "epoch": 17.16020775069916,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 42952
    },
    {
      "epoch": 17.160607271274472,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 42953
    },
    {
      "epoch": 17.16100679184978,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 42954
    },
    {
      "epoch": 17.16140631242509,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 42955
    },
    {
      "epoch": 17.1618058330004,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 42956
    },
    {
      "epoch": 17.162205353575708,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 42957
    },
    {
      "epoch": 17.162604874151018,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 42958
    },
    {
      "epoch": 17.16300439472633,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 42959
    },
    {
      "epoch": 17.16340391530164,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 42960
    },
    {
      "epoch": 17.163803435876947,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 42961
    },
    {
      "epoch": 17.164202956452257,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 42962
    },
    {
      "epoch": 17.164602477027568,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 42963
    },
    {
      "epoch": 17.165001997602875,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 42964
    },
    {
      "epoch": 17.165401518178186,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 42965
    },
    {
      "epoch": 17.165801038753496,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 42966
    },
    {
      "epoch": 17.166200559328807,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 42967
    },
    {
      "epoch": 17.166600079904114,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 42968
    },
    {
      "epoch": 17.166999600479425,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 42969
    },
    {
      "epoch": 17.167399121054736,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 42970
    },
    {
      "epoch": 17.167798641630043,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 42971
    },
    {
      "epoch": 17.168198162205353,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 42972
    },
    {
      "epoch": 17.168597682780664,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 42973
    },
    {
      "epoch": 17.16899720335597,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 42974
    },
    {
      "epoch": 17.169396723931282,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 42975
    },
    {
      "epoch": 17.169796244506593,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 42976
    },
    {
      "epoch": 17.170195765081903,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 42977
    },
    {
      "epoch": 17.17059528565721,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 42978
    },
    {
      "epoch": 17.17099480623252,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 42979
    },
    {
      "epoch": 17.17139432680783,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 42980
    },
    {
      "epoch": 17.17179384738314,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 42981
    },
    {
      "epoch": 17.17219336795845,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 42982
    },
    {
      "epoch": 17.17259288853376,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 42983
    },
    {
      "epoch": 17.17299240910907,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 42984
    },
    {
      "epoch": 17.173391929684378,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 42985
    },
    {
      "epoch": 17.17379145025969,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 42986
    },
    {
      "epoch": 17.174190970835,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 42987
    },
    {
      "epoch": 17.174590491410306,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 42988
    },
    {
      "epoch": 17.174990011985617,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 42989
    },
    {
      "epoch": 17.175389532560928,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 42990
    },
    {
      "epoch": 17.175789053136235,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 42991
    },
    {
      "epoch": 17.176188573711546,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 42992
    },
    {
      "epoch": 17.176588094286856,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 42993
    },
    {
      "epoch": 17.176987614862167,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 42994
    },
    {
      "epoch": 17.177387135437474,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 42995
    },
    {
      "epoch": 17.177786656012785,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 42996
    },
    {
      "epoch": 17.178186176588095,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 42997
    },
    {
      "epoch": 17.178585697163403,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 42998
    },
    {
      "epoch": 17.178985217738713,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 42999
    },
    {
      "epoch": 17.179384738314024,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 43000
    },
    {
      "epoch": 17.179784258889335,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 43001
    },
    {
      "epoch": 17.18018377946464,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 43002
    },
    {
      "epoch": 17.180583300039952,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 43003
    },
    {
      "epoch": 17.180982820615263,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0101,
      "step": 43004
    },
    {
      "epoch": 17.18138234119057,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 43005
    },
    {
      "epoch": 17.18178186176588,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 43006
    },
    {
      "epoch": 17.18218138234119,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 43007
    },
    {
      "epoch": 17.1825809029165,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 43008
    },
    {
      "epoch": 17.18298042349181,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 43009
    },
    {
      "epoch": 17.18337994406712,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 43010
    },
    {
      "epoch": 17.18377946464243,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 43011
    },
    {
      "epoch": 17.184178985217738,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 43012
    },
    {
      "epoch": 17.18457850579305,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 43013
    },
    {
      "epoch": 17.18497802636836,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 43014
    },
    {
      "epoch": 17.185377546943666,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 43015
    },
    {
      "epoch": 17.185777067518977,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 43016
    },
    {
      "epoch": 17.186176588094288,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 43017
    },
    {
      "epoch": 17.186576108669595,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 43018
    },
    {
      "epoch": 17.186975629244905,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 43019
    },
    {
      "epoch": 17.187375149820216,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 43020
    },
    {
      "epoch": 17.187774670395527,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 43021
    },
    {
      "epoch": 17.188174190970834,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 43022
    },
    {
      "epoch": 17.188573711546145,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 43023
    },
    {
      "epoch": 17.188973232121455,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 43024
    },
    {
      "epoch": 17.189372752696762,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 43025
    },
    {
      "epoch": 17.189772273272073,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 43026
    },
    {
      "epoch": 17.190171793847384,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 43027
    },
    {
      "epoch": 17.190571314422694,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 43028
    },
    {
      "epoch": 17.190970834998,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 43029
    },
    {
      "epoch": 17.191370355573312,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 43030
    },
    {
      "epoch": 17.191769876148623,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 43031
    },
    {
      "epoch": 17.19216939672393,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 43032
    },
    {
      "epoch": 17.19256891729924,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 43033
    },
    {
      "epoch": 17.19296843787455,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 43034
    },
    {
      "epoch": 17.19336795844986,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 43035
    },
    {
      "epoch": 17.19376747902517,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 43036
    },
    {
      "epoch": 17.19416699960048,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 43037
    },
    {
      "epoch": 17.19456652017579,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 43038
    },
    {
      "epoch": 17.194966040751098,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0116,
      "step": 43039
    },
    {
      "epoch": 17.19536556132641,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 43040
    },
    {
      "epoch": 17.19576508190172,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 43041
    },
    {
      "epoch": 17.196164602477026,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 43042
    },
    {
      "epoch": 17.196564123052337,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 43043
    },
    {
      "epoch": 17.196963643627647,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 43044
    },
    {
      "epoch": 17.197363164202958,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 43045
    },
    {
      "epoch": 17.197762684778265,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0069,
      "step": 43046
    },
    {
      "epoch": 17.198162205353576,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 43047
    },
    {
      "epoch": 17.198561725928887,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 43048
    },
    {
      "epoch": 17.198961246504194,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 43049
    },
    {
      "epoch": 17.199360767079504,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 43050
    },
    {
      "epoch": 17.199760287654815,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 43051
    },
    {
      "epoch": 17.200159808230122,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 43052
    },
    {
      "epoch": 17.200559328805433,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 43053
    },
    {
      "epoch": 17.200958849380743,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 43054
    },
    {
      "epoch": 17.201358369956054,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 43055
    },
    {
      "epoch": 17.20175789053136,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 43056
    },
    {
      "epoch": 17.202157411106672,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 43057
    },
    {
      "epoch": 17.202556931681983,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 43058
    },
    {
      "epoch": 17.20295645225729,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 43059
    },
    {
      "epoch": 17.2033559728326,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 43060
    },
    {
      "epoch": 17.20375549340791,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 43061
    },
    {
      "epoch": 17.204155013983222,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 43062
    },
    {
      "epoch": 17.20455453455853,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 43063
    },
    {
      "epoch": 17.20495405513384,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 43064
    },
    {
      "epoch": 17.20535357570915,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 43065
    },
    {
      "epoch": 17.205753096284457,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 43066
    },
    {
      "epoch": 17.206152616859768,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 43067
    },
    {
      "epoch": 17.20655213743508,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 43068
    },
    {
      "epoch": 17.206951658010386,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 43069
    },
    {
      "epoch": 17.207351178585697,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 43070
    },
    {
      "epoch": 17.207750699161007,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 43071
    },
    {
      "epoch": 17.208150219736318,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 43072
    },
    {
      "epoch": 17.208549740311625,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 43073
    },
    {
      "epoch": 17.208949260886936,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 43074
    },
    {
      "epoch": 17.209348781462246,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 43075
    },
    {
      "epoch": 17.209748302037553,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 43076
    },
    {
      "epoch": 17.210147822612864,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 43077
    },
    {
      "epoch": 17.210547343188175,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 43078
    },
    {
      "epoch": 17.210946863763485,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 43079
    },
    {
      "epoch": 17.211346384338793,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 43080
    },
    {
      "epoch": 17.211745904914103,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 43081
    },
    {
      "epoch": 17.212145425489414,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 43082
    },
    {
      "epoch": 17.21254494606472,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.012,
      "step": 43083
    },
    {
      "epoch": 17.21294446664003,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 43084
    },
    {
      "epoch": 17.213343987215342,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 43085
    },
    {
      "epoch": 17.21374350779065,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 43086
    },
    {
      "epoch": 17.21414302836596,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 43087
    },
    {
      "epoch": 17.21454254894127,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 43088
    },
    {
      "epoch": 17.21494206951658,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 43089
    },
    {
      "epoch": 17.21534159009189,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 43090
    },
    {
      "epoch": 17.2157411106672,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 43091
    },
    {
      "epoch": 17.21614063124251,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 43092
    },
    {
      "epoch": 17.216540151817817,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 43093
    },
    {
      "epoch": 17.216939672393128,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 43094
    },
    {
      "epoch": 17.21733919296844,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 43095
    },
    {
      "epoch": 17.21773871354375,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 43096
    },
    {
      "epoch": 17.218138234119056,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 43097
    },
    {
      "epoch": 17.218537754694367,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 43098
    },
    {
      "epoch": 17.218937275269678,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 43099
    },
    {
      "epoch": 17.219336795844985,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 43100
    },
    {
      "epoch": 17.219736316420295,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 43101
    },
    {
      "epoch": 17.220135836995606,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 43102
    },
    {
      "epoch": 17.220535357570913,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 43103
    },
    {
      "epoch": 17.220934878146224,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 43104
    },
    {
      "epoch": 17.221334398721535,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 43105
    },
    {
      "epoch": 17.221733919296845,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 43106
    },
    {
      "epoch": 17.222133439872152,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 43107
    },
    {
      "epoch": 17.222532960447463,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 43108
    },
    {
      "epoch": 17.222932481022774,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 43109
    },
    {
      "epoch": 17.22333200159808,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 43110
    },
    {
      "epoch": 17.22373152217339,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 43111
    },
    {
      "epoch": 17.224131042748702,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 43112
    },
    {
      "epoch": 17.224530563324013,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 43113
    },
    {
      "epoch": 17.22493008389932,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 43114
    },
    {
      "epoch": 17.22532960447463,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 43115
    },
    {
      "epoch": 17.22572912504994,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 43116
    },
    {
      "epoch": 17.22612864562525,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 43117
    },
    {
      "epoch": 17.22652816620056,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 43118
    },
    {
      "epoch": 17.22692768677587,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 43119
    },
    {
      "epoch": 17.227327207351177,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 43120
    },
    {
      "epoch": 17.227726727926488,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 43121
    },
    {
      "epoch": 17.2281262485018,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 43122
    },
    {
      "epoch": 17.22852576907711,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 43123
    },
    {
      "epoch": 17.228925289652416,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 43124
    },
    {
      "epoch": 17.229324810227727,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0063,
      "step": 43125
    },
    {
      "epoch": 17.229724330803037,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 43126
    },
    {
      "epoch": 17.230123851378345,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 43127
    },
    {
      "epoch": 17.230523371953655,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 43128
    },
    {
      "epoch": 17.230922892528966,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 43129
    },
    {
      "epoch": 17.231322413104277,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 43130
    },
    {
      "epoch": 17.231721933679584,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 43131
    },
    {
      "epoch": 17.232121454254894,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 43132
    },
    {
      "epoch": 17.232520974830205,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 43133
    },
    {
      "epoch": 17.232920495405512,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 43134
    },
    {
      "epoch": 17.233320015980823,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 43135
    },
    {
      "epoch": 17.233719536556134,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 43136
    },
    {
      "epoch": 17.23411905713144,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 43137
    },
    {
      "epoch": 17.23451857770675,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 43138
    },
    {
      "epoch": 17.234918098282062,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 43139
    },
    {
      "epoch": 17.235317618857373,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 43140
    },
    {
      "epoch": 17.23571713943268,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 43141
    },
    {
      "epoch": 17.23611666000799,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 43142
    },
    {
      "epoch": 17.2365161805833,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 43143
    },
    {
      "epoch": 17.23691570115861,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 43144
    },
    {
      "epoch": 17.23731522173392,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 43145
    },
    {
      "epoch": 17.23771474230923,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 43146
    },
    {
      "epoch": 17.23811426288454,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 43147
    },
    {
      "epoch": 17.238513783459847,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 43148
    },
    {
      "epoch": 17.238913304035158,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 43149
    },
    {
      "epoch": 17.23931282461047,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 43150
    },
    {
      "epoch": 17.239712345185776,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 43151
    },
    {
      "epoch": 17.240111865761087,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 43152
    },
    {
      "epoch": 17.240511386336397,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0074,
      "step": 43153
    },
    {
      "epoch": 17.240910906911704,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 43154
    },
    {
      "epoch": 17.241310427487015,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 43155
    },
    {
      "epoch": 17.241709948062326,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 43156
    },
    {
      "epoch": 17.242109468637636,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 43157
    },
    {
      "epoch": 17.242508989212944,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 43158
    },
    {
      "epoch": 17.242908509788254,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 43159
    },
    {
      "epoch": 17.243308030363565,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 43160
    },
    {
      "epoch": 17.243707550938872,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 43161
    },
    {
      "epoch": 17.244107071514183,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 43162
    },
    {
      "epoch": 17.244506592089493,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 43163
    },
    {
      "epoch": 17.244906112664804,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 43164
    },
    {
      "epoch": 17.24530563324011,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 43165
    },
    {
      "epoch": 17.245705153815422,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 43166
    },
    {
      "epoch": 17.246104674390732,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 43167
    },
    {
      "epoch": 17.24650419496604,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 43168
    },
    {
      "epoch": 17.24690371554135,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 43169
    },
    {
      "epoch": 17.24730323611666,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 43170
    },
    {
      "epoch": 17.247702756691968,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0096,
      "step": 43171
    },
    {
      "epoch": 17.24810227726728,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 43172
    },
    {
      "epoch": 17.24850179784259,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0103,
      "step": 43173
    },
    {
      "epoch": 17.2489013184179,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 43174
    },
    {
      "epoch": 17.249300838993207,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 43175
    },
    {
      "epoch": 17.249700359568518,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0134,
      "step": 43176
    },
    {
      "epoch": 17.25009988014383,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 43177
    },
    {
      "epoch": 17.250499400719136,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 43178
    },
    {
      "epoch": 17.250898921294446,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 43179
    },
    {
      "epoch": 17.251298441869757,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 43180
    },
    {
      "epoch": 17.251697962445064,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 43181
    },
    {
      "epoch": 17.252097483020375,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 43182
    },
    {
      "epoch": 17.252497003595685,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0065,
      "step": 43183
    },
    {
      "epoch": 17.252896524170996,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 43184
    },
    {
      "epoch": 17.253296044746303,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 43185
    },
    {
      "epoch": 17.253695565321614,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 43186
    },
    {
      "epoch": 17.254095085896925,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 43187
    },
    {
      "epoch": 17.25449460647223,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 43188
    },
    {
      "epoch": 17.254894127047542,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 43189
    },
    {
      "epoch": 17.255293647622853,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 43190
    },
    {
      "epoch": 17.255693168198164,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 43191
    },
    {
      "epoch": 17.25609268877347,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 43192
    },
    {
      "epoch": 17.25649220934878,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 43193
    },
    {
      "epoch": 17.256891729924092,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 43194
    },
    {
      "epoch": 17.2572912504994,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 43195
    },
    {
      "epoch": 17.25769077107471,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 43196
    },
    {
      "epoch": 17.25809029165002,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 43197
    },
    {
      "epoch": 17.258489812225328,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 43198
    },
    {
      "epoch": 17.25888933280064,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 43199
    },
    {
      "epoch": 17.25928885337595,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 43200
    },
    {
      "epoch": 17.25968837395126,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 43201
    },
    {
      "epoch": 17.260087894526567,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 43202
    },
    {
      "epoch": 17.260487415101878,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 43203
    },
    {
      "epoch": 17.26088693567719,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 43204
    },
    {
      "epoch": 17.261286456252495,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 43205
    },
    {
      "epoch": 17.261685976827806,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 43206
    },
    {
      "epoch": 17.262085497403117,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 43207
    },
    {
      "epoch": 17.262485017978427,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 43208
    },
    {
      "epoch": 17.262884538553735,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 43209
    },
    {
      "epoch": 17.263284059129045,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 43210
    },
    {
      "epoch": 17.263683579704356,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 43211
    },
    {
      "epoch": 17.264083100279663,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 43212
    },
    {
      "epoch": 17.264482620854974,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 43213
    },
    {
      "epoch": 17.264882141430284,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 43214
    },
    {
      "epoch": 17.26528166200559,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 43215
    },
    {
      "epoch": 17.265681182580902,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 43216
    },
    {
      "epoch": 17.266080703156213,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 43217
    },
    {
      "epoch": 17.266480223731524,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 43218
    },
    {
      "epoch": 17.26687974430683,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 43219
    },
    {
      "epoch": 17.26727926488214,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 43220
    },
    {
      "epoch": 17.267678785457452,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 43221
    },
    {
      "epoch": 17.26807830603276,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 43222
    },
    {
      "epoch": 17.26847782660807,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 43223
    },
    {
      "epoch": 17.26887734718338,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 43224
    },
    {
      "epoch": 17.26927686775869,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 43225
    },
    {
      "epoch": 17.269676388334,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 43226
    },
    {
      "epoch": 17.27007590890931,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 43227
    },
    {
      "epoch": 17.27047542948462,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 43228
    },
    {
      "epoch": 17.270874950059927,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 43229
    },
    {
      "epoch": 17.271274470635237,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 43230
    },
    {
      "epoch": 17.271673991210548,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 43231
    },
    {
      "epoch": 17.272073511785855,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 43232
    },
    {
      "epoch": 17.272473032361166,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 43233
    },
    {
      "epoch": 17.272872552936477,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 43234
    },
    {
      "epoch": 17.273272073511787,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 43235
    },
    {
      "epoch": 17.273671594087094,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 43236
    },
    {
      "epoch": 17.274071114662405,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 43237
    },
    {
      "epoch": 17.274470635237716,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 43238
    },
    {
      "epoch": 17.274870155813023,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 43239
    },
    {
      "epoch": 17.275269676388334,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 43240
    },
    {
      "epoch": 17.275669196963644,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 43241
    },
    {
      "epoch": 17.276068717538955,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 43242
    },
    {
      "epoch": 17.276468238114262,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 43243
    },
    {
      "epoch": 17.276867758689573,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 43244
    },
    {
      "epoch": 17.277267279264883,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 43245
    },
    {
      "epoch": 17.27766679984019,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 43246
    },
    {
      "epoch": 17.2780663204155,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 43247
    },
    {
      "epoch": 17.278465840990812,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 43248
    },
    {
      "epoch": 17.27886536156612,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 43249
    },
    {
      "epoch": 17.27926488214143,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 43250
    },
    {
      "epoch": 17.27966440271674,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 43251
    },
    {
      "epoch": 17.28006392329205,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 43252
    },
    {
      "epoch": 17.280463443867358,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 43253
    },
    {
      "epoch": 17.28086296444267,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 43254
    },
    {
      "epoch": 17.28126248501798,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 43255
    },
    {
      "epoch": 17.281662005593287,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 43256
    },
    {
      "epoch": 17.282061526168597,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 43257
    },
    {
      "epoch": 17.282461046743908,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 43258
    },
    {
      "epoch": 17.28286056731922,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 43259
    },
    {
      "epoch": 17.283260087894526,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 43260
    },
    {
      "epoch": 17.283659608469836,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 43261
    },
    {
      "epoch": 17.284059129045147,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 43262
    },
    {
      "epoch": 17.284458649620454,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 43263
    },
    {
      "epoch": 17.284858170195765,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 43264
    },
    {
      "epoch": 17.285257690771076,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 43265
    },
    {
      "epoch": 17.285657211346383,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 43266
    },
    {
      "epoch": 17.286056731921693,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 43267
    },
    {
      "epoch": 17.286456252497004,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 43268
    },
    {
      "epoch": 17.286855773072315,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 43269
    },
    {
      "epoch": 17.287255293647622,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 43270
    },
    {
      "epoch": 17.287654814222932,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0084,
      "step": 43271
    },
    {
      "epoch": 17.288054334798243,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 43272
    },
    {
      "epoch": 17.28845385537355,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 43273
    },
    {
      "epoch": 17.28885337594886,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 43274
    },
    {
      "epoch": 17.28925289652417,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 43275
    },
    {
      "epoch": 17.289652417099482,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 43276
    },
    {
      "epoch": 17.29005193767479,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 43277
    },
    {
      "epoch": 17.2904514582501,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0115,
      "step": 43278
    },
    {
      "epoch": 17.29085097882541,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 43279
    },
    {
      "epoch": 17.291250499400718,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0126,
      "step": 43280
    },
    {
      "epoch": 17.29165001997603,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0053,
      "step": 43281
    },
    {
      "epoch": 17.29204954055134,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 43282
    },
    {
      "epoch": 17.292449061126646,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 43283
    },
    {
      "epoch": 17.292848581701957,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 43284
    },
    {
      "epoch": 17.293248102277268,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 43285
    },
    {
      "epoch": 17.29364762285258,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 43286
    },
    {
      "epoch": 17.294047143427886,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 43287
    },
    {
      "epoch": 17.294446664003196,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 43288
    },
    {
      "epoch": 17.294846184578507,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 43289
    },
    {
      "epoch": 17.295245705153814,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 43290
    },
    {
      "epoch": 17.295645225729125,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 43291
    },
    {
      "epoch": 17.296044746304435,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 43292
    },
    {
      "epoch": 17.296444266879746,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 43293
    },
    {
      "epoch": 17.296843787455053,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 43294
    },
    {
      "epoch": 17.297243308030364,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 43295
    },
    {
      "epoch": 17.297642828605674,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 43296
    },
    {
      "epoch": 17.29804234918098,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 43297
    },
    {
      "epoch": 17.298441869756292,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 43298
    },
    {
      "epoch": 17.298841390331603,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 43299
    },
    {
      "epoch": 17.29924091090691,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 43300
    },
    {
      "epoch": 17.29964043148222,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 43301
    },
    {
      "epoch": 17.30003995205753,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 43302
    },
    {
      "epoch": 17.300439472632842,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 43303
    },
    {
      "epoch": 17.30083899320815,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 43304
    },
    {
      "epoch": 17.30123851378346,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 43305
    },
    {
      "epoch": 17.30163803435877,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 43306
    },
    {
      "epoch": 17.302037554934078,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 43307
    },
    {
      "epoch": 17.30243707550939,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 43308
    },
    {
      "epoch": 17.3028365960847,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 43309
    },
    {
      "epoch": 17.30323611666001,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 43310
    },
    {
      "epoch": 17.303635637235317,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 43311
    },
    {
      "epoch": 17.304035157810628,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 43312
    },
    {
      "epoch": 17.304434678385938,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 43313
    },
    {
      "epoch": 17.304834198961245,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 43314
    },
    {
      "epoch": 17.305233719536556,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 43315
    },
    {
      "epoch": 17.305633240111867,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 43316
    },
    {
      "epoch": 17.306032760687174,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 43317
    },
    {
      "epoch": 17.306432281262484,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 43318
    },
    {
      "epoch": 17.306831801837795,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 43319
    },
    {
      "epoch": 17.307231322413106,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 43320
    },
    {
      "epoch": 17.307630842988413,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 43321
    },
    {
      "epoch": 17.308030363563724,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 43322
    },
    {
      "epoch": 17.308429884139034,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 43323
    },
    {
      "epoch": 17.30882940471434,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 43324
    },
    {
      "epoch": 17.309228925289652,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 43325
    },
    {
      "epoch": 17.309628445864963,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 43326
    },
    {
      "epoch": 17.310027966440273,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 43327
    },
    {
      "epoch": 17.31042748701558,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0103,
      "step": 43328
    },
    {
      "epoch": 17.31082700759089,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0491,
      "step": 43329
    },
    {
      "epoch": 17.311226528166202,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 43330
    },
    {
      "epoch": 17.31162604874151,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 43331
    },
    {
      "epoch": 17.31202556931682,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 43332
    },
    {
      "epoch": 17.31242508989213,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 43333
    },
    {
      "epoch": 17.312824610467437,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 43334
    },
    {
      "epoch": 17.313224131042748,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0125,
      "step": 43335
    },
    {
      "epoch": 17.31362365161806,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 43336
    },
    {
      "epoch": 17.31402317219337,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 43337
    },
    {
      "epoch": 17.314422692768677,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 43338
    },
    {
      "epoch": 17.314822213343987,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 43339
    },
    {
      "epoch": 17.315221733919298,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0117,
      "step": 43340
    },
    {
      "epoch": 17.315621254494605,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 43341
    },
    {
      "epoch": 17.316020775069916,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 43342
    },
    {
      "epoch": 17.316420295645226,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 43343
    },
    {
      "epoch": 17.316819816220537,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 43344
    },
    {
      "epoch": 17.317219336795844,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 43345
    },
    {
      "epoch": 17.317618857371155,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 43346
    },
    {
      "epoch": 17.318018377946466,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 43347
    },
    {
      "epoch": 17.318417898521773,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 43348
    },
    {
      "epoch": 17.318817419097083,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 43349
    },
    {
      "epoch": 17.319216939672394,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 43350
    },
    {
      "epoch": 17.3196164602477,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 43351
    },
    {
      "epoch": 17.320015980823012,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 43352
    },
    {
      "epoch": 17.320415501398323,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 43353
    },
    {
      "epoch": 17.320815021973633,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 43354
    },
    {
      "epoch": 17.32121454254894,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 43355
    },
    {
      "epoch": 17.32161406312425,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 43356
    },
    {
      "epoch": 17.32201358369956,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 43357
    },
    {
      "epoch": 17.32241310427487,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 43358
    },
    {
      "epoch": 17.32281262485018,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 43359
    },
    {
      "epoch": 17.32321214542549,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 43360
    },
    {
      "epoch": 17.3236116660008,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 43361
    },
    {
      "epoch": 17.324011186576108,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 43362
    },
    {
      "epoch": 17.32441070715142,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 43363
    },
    {
      "epoch": 17.32481022772673,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 43364
    },
    {
      "epoch": 17.325209748302036,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 43365
    },
    {
      "epoch": 17.325609268877347,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 43366
    },
    {
      "epoch": 17.326008789452658,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 43367
    },
    {
      "epoch": 17.326408310027965,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 43368
    },
    {
      "epoch": 17.326807830603276,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 43369
    },
    {
      "epoch": 17.327207351178586,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 43370
    },
    {
      "epoch": 17.327606871753897,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 43371
    },
    {
      "epoch": 17.328006392329204,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 43372
    },
    {
      "epoch": 17.328405912904515,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 43373
    },
    {
      "epoch": 17.328805433479825,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 43374
    },
    {
      "epoch": 17.329204954055133,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 43375
    },
    {
      "epoch": 17.329604474630443,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 43376
    },
    {
      "epoch": 17.330003995205754,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 43377
    },
    {
      "epoch": 17.330403515781065,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 43378
    },
    {
      "epoch": 17.33080303635637,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 43379
    },
    {
      "epoch": 17.331202556931682,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 43380
    },
    {
      "epoch": 17.331602077506993,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 43381
    },
    {
      "epoch": 17.3320015980823,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 43382
    },
    {
      "epoch": 17.33240111865761,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 43383
    },
    {
      "epoch": 17.33280063923292,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 43384
    },
    {
      "epoch": 17.33320015980823,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 43385
    },
    {
      "epoch": 17.33359968038354,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 43386
    },
    {
      "epoch": 17.33399920095885,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 43387
    },
    {
      "epoch": 17.33439872153416,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 43388
    },
    {
      "epoch": 17.334798242109468,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 43389
    },
    {
      "epoch": 17.33519776268478,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 43390
    },
    {
      "epoch": 17.33559728326009,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 43391
    },
    {
      "epoch": 17.335996803835396,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 43392
    },
    {
      "epoch": 17.336396324410707,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 43393
    },
    {
      "epoch": 17.336795844986018,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 43394
    },
    {
      "epoch": 17.337195365561325,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 43395
    },
    {
      "epoch": 17.337594886136635,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 43396
    },
    {
      "epoch": 17.337994406711946,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 43397
    },
    {
      "epoch": 17.338393927287257,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 43398
    },
    {
      "epoch": 17.338793447862564,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 43399
    },
    {
      "epoch": 17.339192968437874,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 43400
    },
    {
      "epoch": 17.339592489013185,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 43401
    },
    {
      "epoch": 17.339992009588492,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 43402
    },
    {
      "epoch": 17.340391530163803,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 43403
    },
    {
      "epoch": 17.340791050739114,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 43404
    },
    {
      "epoch": 17.341190571314424,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 43405
    },
    {
      "epoch": 17.34159009188973,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 43406
    },
    {
      "epoch": 17.341989612465042,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 43407
    },
    {
      "epoch": 17.342389133040353,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 43408
    },
    {
      "epoch": 17.34278865361566,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 43409
    },
    {
      "epoch": 17.34318817419097,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 43410
    },
    {
      "epoch": 17.34358769476628,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 43411
    },
    {
      "epoch": 17.34398721534159,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 43412
    },
    {
      "epoch": 17.3443867359169,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 43413
    },
    {
      "epoch": 17.34478625649221,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0102,
      "step": 43414
    },
    {
      "epoch": 17.34518577706752,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 43415
    },
    {
      "epoch": 17.345585297642828,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 43416
    },
    {
      "epoch": 17.345984818218138,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 43417
    },
    {
      "epoch": 17.34638433879345,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 43418
    },
    {
      "epoch": 17.346783859368756,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0075,
      "step": 43419
    },
    {
      "epoch": 17.347183379944067,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 43420
    },
    {
      "epoch": 17.347582900519377,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 43421
    },
    {
      "epoch": 17.347982421094688,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 43422
    },
    {
      "epoch": 17.348381941669995,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 43423
    },
    {
      "epoch": 17.348781462245306,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 43424
    },
    {
      "epoch": 17.349180982820616,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 43425
    },
    {
      "epoch": 17.349580503395924,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 43426
    },
    {
      "epoch": 17.349980023971234,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 43427
    },
    {
      "epoch": 17.350379544546545,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 43428
    },
    {
      "epoch": 17.350779065121852,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0094,
      "step": 43429
    },
    {
      "epoch": 17.351178585697163,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 43430
    },
    {
      "epoch": 17.351578106272473,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 43431
    },
    {
      "epoch": 17.351977626847784,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 43432
    },
    {
      "epoch": 17.35237714742309,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 43433
    },
    {
      "epoch": 17.352776667998402,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 43434
    },
    {
      "epoch": 17.353176188573713,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 43435
    },
    {
      "epoch": 17.35357570914902,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 43436
    },
    {
      "epoch": 17.35397522972433,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 43437
    },
    {
      "epoch": 17.35437475029964,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 43438
    },
    {
      "epoch": 17.35477427087495,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 43439
    },
    {
      "epoch": 17.35517379145026,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 43440
    },
    {
      "epoch": 17.35557331202557,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 43441
    },
    {
      "epoch": 17.35597283260088,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 43442
    },
    {
      "epoch": 17.356372353176187,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 43443
    },
    {
      "epoch": 17.356771873751498,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 43444
    },
    {
      "epoch": 17.35717139432681,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 43445
    },
    {
      "epoch": 17.357570914902116,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 43446
    },
    {
      "epoch": 17.357970435477426,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 43447
    },
    {
      "epoch": 17.358369956052737,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 43448
    },
    {
      "epoch": 17.358769476628048,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 43449
    },
    {
      "epoch": 17.359168997203355,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 43450
    },
    {
      "epoch": 17.359568517778666,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 43451
    },
    {
      "epoch": 17.359968038353976,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 43452
    },
    {
      "epoch": 17.360367558929283,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 43453
    },
    {
      "epoch": 17.360767079504594,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 43454
    },
    {
      "epoch": 17.361166600079905,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 43455
    },
    {
      "epoch": 17.361566120655215,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 43456
    },
    {
      "epoch": 17.361965641230523,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 43457
    },
    {
      "epoch": 17.362365161805833,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 43458
    },
    {
      "epoch": 17.362764682381144,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0385,
      "step": 43459
    },
    {
      "epoch": 17.36316420295645,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 43460
    },
    {
      "epoch": 17.36356372353176,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.01,
      "step": 43461
    },
    {
      "epoch": 17.363963244107072,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 43462
    },
    {
      "epoch": 17.36436276468238,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 43463
    },
    {
      "epoch": 17.36476228525769,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 43464
    },
    {
      "epoch": 17.365161805833,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 43465
    },
    {
      "epoch": 17.36556132640831,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 43466
    },
    {
      "epoch": 17.36596084698362,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 43467
    },
    {
      "epoch": 17.36636036755893,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 43468
    },
    {
      "epoch": 17.36675988813424,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 43469
    },
    {
      "epoch": 17.367159408709547,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 43470
    },
    {
      "epoch": 17.367558929284858,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 43471
    },
    {
      "epoch": 17.36795844986017,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 43472
    },
    {
      "epoch": 17.36835797043548,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 43473
    },
    {
      "epoch": 17.368757491010786,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 43474
    },
    {
      "epoch": 17.369157011586097,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 43475
    },
    {
      "epoch": 17.369556532161408,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 43476
    },
    {
      "epoch": 17.369956052736715,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 43477
    },
    {
      "epoch": 17.370355573312025,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 43478
    },
    {
      "epoch": 17.370755093887336,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 43479
    },
    {
      "epoch": 17.371154614462643,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 43480
    },
    {
      "epoch": 17.371554135037954,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 43481
    },
    {
      "epoch": 17.371953655613265,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 43482
    },
    {
      "epoch": 17.372353176188575,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 43483
    },
    {
      "epoch": 17.372752696763882,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 43484
    },
    {
      "epoch": 17.373152217339193,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 43485
    },
    {
      "epoch": 17.373551737914504,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 43486
    },
    {
      "epoch": 17.37395125848981,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 43487
    },
    {
      "epoch": 17.37435077906512,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 43488
    },
    {
      "epoch": 17.374750299640432,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 43489
    },
    {
      "epoch": 17.375149820215743,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 43490
    },
    {
      "epoch": 17.37554934079105,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 43491
    },
    {
      "epoch": 17.37594886136636,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 43492
    },
    {
      "epoch": 17.37634838194167,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 43493
    },
    {
      "epoch": 17.37674790251698,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 43494
    },
    {
      "epoch": 17.37714742309229,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 43495
    },
    {
      "epoch": 17.3775469436676,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 43496
    },
    {
      "epoch": 17.377946464242907,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 43497
    },
    {
      "epoch": 17.378345984818218,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 43498
    },
    {
      "epoch": 17.37874550539353,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 43499
    },
    {
      "epoch": 17.37914502596884,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 43500
    },
    {
      "epoch": 17.379544546544146,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0111,
      "step": 43501
    },
    {
      "epoch": 17.379944067119457,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 43502
    },
    {
      "epoch": 17.380343587694767,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 43503
    },
    {
      "epoch": 17.380743108270075,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 43504
    },
    {
      "epoch": 17.381142628845385,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 43505
    },
    {
      "epoch": 17.381542149420696,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 43506
    },
    {
      "epoch": 17.381941669996007,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 43507
    },
    {
      "epoch": 17.382341190571314,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 43508
    },
    {
      "epoch": 17.382740711146624,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 43509
    },
    {
      "epoch": 17.383140231721935,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 43510
    },
    {
      "epoch": 17.383539752297242,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 43511
    },
    {
      "epoch": 17.383939272872553,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 43512
    },
    {
      "epoch": 17.384338793447863,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 43513
    },
    {
      "epoch": 17.38473831402317,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 43514
    },
    {
      "epoch": 17.38513783459848,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 43515
    },
    {
      "epoch": 17.385537355173792,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 43516
    },
    {
      "epoch": 17.385936875749103,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 43517
    },
    {
      "epoch": 17.38633639632441,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 43518
    },
    {
      "epoch": 17.38673591689972,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 43519
    },
    {
      "epoch": 17.38713543747503,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 43520
    },
    {
      "epoch": 17.387534958050338,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 43521
    },
    {
      "epoch": 17.38793447862565,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 43522
    },
    {
      "epoch": 17.38833399920096,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 43523
    },
    {
      "epoch": 17.388733519776267,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 43524
    },
    {
      "epoch": 17.389133040351577,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 43525
    },
    {
      "epoch": 17.389532560926888,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 43526
    },
    {
      "epoch": 17.3899320815022,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 43527
    },
    {
      "epoch": 17.390331602077506,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 43528
    },
    {
      "epoch": 17.390731122652817,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 43529
    },
    {
      "epoch": 17.391130643228127,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 43530
    },
    {
      "epoch": 17.391530163803434,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 43531
    },
    {
      "epoch": 17.391929684378745,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 43532
    },
    {
      "epoch": 17.392329204954056,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 43533
    },
    {
      "epoch": 17.392728725529366,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 43534
    },
    {
      "epoch": 17.393128246104673,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 43535
    },
    {
      "epoch": 17.393527766679984,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 43536
    },
    {
      "epoch": 17.393927287255295,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 43537
    },
    {
      "epoch": 17.394326807830602,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 43538
    },
    {
      "epoch": 17.394726328405913,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0005,
      "loss": 1.04,
      "step": 43539
    },
    {
      "epoch": 17.395125848981223,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 43540
    },
    {
      "epoch": 17.39552536955653,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 43541
    },
    {
      "epoch": 17.39592489013184,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 43542
    },
    {
      "epoch": 17.39632441070715,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 43543
    },
    {
      "epoch": 17.396723931282462,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0025,
      "step": 43544
    },
    {
      "epoch": 17.39712345185777,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 43545
    },
    {
      "epoch": 17.39752297243308,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 43546
    },
    {
      "epoch": 17.39792249300839,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 43547
    },
    {
      "epoch": 17.398322013583698,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 43548
    },
    {
      "epoch": 17.39872153415901,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 43549
    },
    {
      "epoch": 17.39912105473432,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 43550
    },
    {
      "epoch": 17.39952057530963,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0105,
      "step": 43551
    },
    {
      "epoch": 17.399920095884937,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 43552
    },
    {
      "epoch": 17.400319616460248,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 43553
    },
    {
      "epoch": 17.40071913703556,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 43554
    },
    {
      "epoch": 17.401118657610866,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 43555
    },
    {
      "epoch": 17.401518178186176,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 43556
    },
    {
      "epoch": 17.401917698761487,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 43557
    },
    {
      "epoch": 17.402317219336794,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 43558
    },
    {
      "epoch": 17.402716739912105,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 43559
    },
    {
      "epoch": 17.403116260487415,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 43560
    },
    {
      "epoch": 17.403515781062726,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 43561
    },
    {
      "epoch": 17.403915301638033,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 43562
    },
    {
      "epoch": 17.404314822213344,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 43563
    },
    {
      "epoch": 17.404714342788655,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 43564
    },
    {
      "epoch": 17.40511386336396,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 43565
    },
    {
      "epoch": 17.405513383939272,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 43566
    },
    {
      "epoch": 17.405912904514583,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 43567
    },
    {
      "epoch": 17.406312425089894,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 43568
    },
    {
      "epoch": 17.4067119456652,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 43569
    },
    {
      "epoch": 17.40711146624051,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 43570
    },
    {
      "epoch": 17.407510986815822,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 43571
    },
    {
      "epoch": 17.40791050739113,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 43572
    },
    {
      "epoch": 17.40831002796644,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 43573
    },
    {
      "epoch": 17.40870954854175,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 43574
    },
    {
      "epoch": 17.409109069117058,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 43575
    },
    {
      "epoch": 17.40950858969237,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 43576
    },
    {
      "epoch": 17.40990811026768,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 43577
    },
    {
      "epoch": 17.41030763084299,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 43578
    },
    {
      "epoch": 17.410707151418297,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 43579
    },
    {
      "epoch": 17.411106671993608,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 43580
    },
    {
      "epoch": 17.41150619256892,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 43581
    },
    {
      "epoch": 17.411905713144225,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 43582
    },
    {
      "epoch": 17.412305233719536,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 43583
    },
    {
      "epoch": 17.412704754294847,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 43584
    },
    {
      "epoch": 17.413104274870157,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 43585
    },
    {
      "epoch": 17.413503795445465,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 43586
    },
    {
      "epoch": 17.413903316020775,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 43587
    },
    {
      "epoch": 17.414302836596086,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 43588
    },
    {
      "epoch": 17.414702357171393,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 43589
    },
    {
      "epoch": 17.415101877746704,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 43590
    },
    {
      "epoch": 17.415501398322014,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 43591
    },
    {
      "epoch": 17.41590091889732,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 43592
    },
    {
      "epoch": 17.416300439472632,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 43593
    },
    {
      "epoch": 17.416699960047943,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 43594
    },
    {
      "epoch": 17.417099480623254,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 43595
    },
    {
      "epoch": 17.41749900119856,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 43596
    },
    {
      "epoch": 17.41789852177387,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0102,
      "step": 43597
    },
    {
      "epoch": 17.418298042349182,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 43598
    },
    {
      "epoch": 17.41869756292449,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 43599
    },
    {
      "epoch": 17.4190970834998,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.009,
      "step": 43600
    },
    {
      "epoch": 17.41949660407511,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 43601
    },
    {
      "epoch": 17.41989612465042,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 43602
    },
    {
      "epoch": 17.42029564522573,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 43603
    },
    {
      "epoch": 17.42069516580104,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0436,
      "step": 43604
    },
    {
      "epoch": 17.42109468637635,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 43605
    },
    {
      "epoch": 17.421494206951657,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 43606
    },
    {
      "epoch": 17.421893727526967,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 43607
    },
    {
      "epoch": 17.422293248102278,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 43608
    },
    {
      "epoch": 17.422692768677585,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 43609
    },
    {
      "epoch": 17.423092289252896,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 43610
    },
    {
      "epoch": 17.423491809828207,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 43611
    },
    {
      "epoch": 17.423891330403517,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 43612
    },
    {
      "epoch": 17.424290850978824,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 43613
    },
    {
      "epoch": 17.424690371554135,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 43614
    },
    {
      "epoch": 17.425089892129446,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 43615
    },
    {
      "epoch": 17.425489412704753,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 43616
    },
    {
      "epoch": 17.425888933280063,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 43617
    },
    {
      "epoch": 17.426288453855374,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 43618
    },
    {
      "epoch": 17.426687974430685,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 43619
    },
    {
      "epoch": 17.427087495005992,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 43620
    },
    {
      "epoch": 17.427487015581303,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 43621
    },
    {
      "epoch": 17.427886536156613,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 43622
    },
    {
      "epoch": 17.42828605673192,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 43623
    },
    {
      "epoch": 17.42868557730723,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 43624
    },
    {
      "epoch": 17.429085097882542,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 43625
    },
    {
      "epoch": 17.42948461845785,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 43626
    },
    {
      "epoch": 17.42988413903316,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 43627
    },
    {
      "epoch": 17.43028365960847,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 43628
    },
    {
      "epoch": 17.43068318018378,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 43629
    },
    {
      "epoch": 17.431082700759088,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 43630
    },
    {
      "epoch": 17.4314822213344,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 43631
    },
    {
      "epoch": 17.43188174190971,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 43632
    },
    {
      "epoch": 17.432281262485017,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 43633
    },
    {
      "epoch": 17.432680783060327,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 43634
    },
    {
      "epoch": 17.433080303635638,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 43635
    },
    {
      "epoch": 17.43347982421095,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 43636
    },
    {
      "epoch": 17.433879344786256,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 43637
    },
    {
      "epoch": 17.434278865361566,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 43638
    },
    {
      "epoch": 17.434678385936877,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 43639
    },
    {
      "epoch": 17.435077906512184,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0121,
      "step": 43640
    },
    {
      "epoch": 17.435477427087495,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 43641
    },
    {
      "epoch": 17.435876947662805,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 43642
    },
    {
      "epoch": 17.436276468238113,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 43643
    },
    {
      "epoch": 17.436675988813423,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 43644
    },
    {
      "epoch": 17.437075509388734,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0448,
      "step": 43645
    },
    {
      "epoch": 17.437475029964045,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 43646
    },
    {
      "epoch": 17.43787455053935,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 43647
    },
    {
      "epoch": 17.438274071114662,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 43648
    },
    {
      "epoch": 17.438673591689973,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 43649
    },
    {
      "epoch": 17.43907311226528,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 43650
    },
    {
      "epoch": 17.43947263284059,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 43651
    },
    {
      "epoch": 17.4398721534159,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 43652
    },
    {
      "epoch": 17.440271673991212,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 43653
    },
    {
      "epoch": 17.44067119456652,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 43654
    },
    {
      "epoch": 17.44107071514183,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 43655
    },
    {
      "epoch": 17.44147023571714,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 43656
    },
    {
      "epoch": 17.441869756292448,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 43657
    },
    {
      "epoch": 17.44226927686776,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 43658
    },
    {
      "epoch": 17.44266879744307,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 43659
    },
    {
      "epoch": 17.443068318018376,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 43660
    },
    {
      "epoch": 17.443467838593687,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 43661
    },
    {
      "epoch": 17.443867359168998,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 43662
    },
    {
      "epoch": 17.44426687974431,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 43663
    },
    {
      "epoch": 17.444666400319615,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 43664
    },
    {
      "epoch": 17.445065920894926,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 43665
    },
    {
      "epoch": 17.445465441470237,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 43666
    },
    {
      "epoch": 17.445864962045544,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 43667
    },
    {
      "epoch": 17.446264482620855,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 43668
    },
    {
      "epoch": 17.446664003196165,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 43669
    },
    {
      "epoch": 17.447063523771476,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 43670
    },
    {
      "epoch": 17.447463044346783,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 43671
    },
    {
      "epoch": 17.447862564922094,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 43672
    },
    {
      "epoch": 17.448262085497404,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 43673
    },
    {
      "epoch": 17.44866160607271,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 43674
    },
    {
      "epoch": 17.449061126648022,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 43675
    },
    {
      "epoch": 17.449460647223333,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 43676
    },
    {
      "epoch": 17.44986016779864,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 43677
    },
    {
      "epoch": 17.45025968837395,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 43678
    },
    {
      "epoch": 17.45065920894926,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 43679
    },
    {
      "epoch": 17.451058729524572,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 43680
    },
    {
      "epoch": 17.45145825009988,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 43681
    },
    {
      "epoch": 17.45185777067519,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0081,
      "step": 43682
    },
    {
      "epoch": 17.4522572912505,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 43683
    },
    {
      "epoch": 17.452656811825808,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 43684
    },
    {
      "epoch": 17.45305633240112,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 43685
    },
    {
      "epoch": 17.45345585297643,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 43686
    },
    {
      "epoch": 17.45385537355174,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 43687
    },
    {
      "epoch": 17.454254894127047,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 43688
    },
    {
      "epoch": 17.454654414702357,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 43689
    },
    {
      "epoch": 17.455053935277668,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 43690
    },
    {
      "epoch": 17.455453455852975,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 43691
    },
    {
      "epoch": 17.455852976428286,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 43692
    },
    {
      "epoch": 17.456252497003597,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 43693
    },
    {
      "epoch": 17.456652017578904,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 43694
    },
    {
      "epoch": 17.457051538154214,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 43695
    },
    {
      "epoch": 17.457451058729525,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 43696
    },
    {
      "epoch": 17.457850579304836,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 43697
    },
    {
      "epoch": 17.458250099880143,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 43698
    },
    {
      "epoch": 17.458649620455454,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 43699
    },
    {
      "epoch": 17.459049141030764,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 43700
    },
    {
      "epoch": 17.45944866160607,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0119,
      "step": 43701
    },
    {
      "epoch": 17.459848182181382,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 43702
    },
    {
      "epoch": 17.460247702756693,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 43703
    },
    {
      "epoch": 17.460647223332003,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 43704
    },
    {
      "epoch": 17.46104674390731,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 43705
    },
    {
      "epoch": 17.46144626448262,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 43706
    },
    {
      "epoch": 17.461845785057932,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 43707
    },
    {
      "epoch": 17.46224530563324,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 43708
    },
    {
      "epoch": 17.46264482620855,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 43709
    },
    {
      "epoch": 17.46304434678386,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 43710
    },
    {
      "epoch": 17.463443867359167,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 43711
    },
    {
      "epoch": 17.463843387934478,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 43712
    },
    {
      "epoch": 17.46424290850979,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 43713
    },
    {
      "epoch": 17.4646424290851,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 43714
    },
    {
      "epoch": 17.465041949660407,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 43715
    },
    {
      "epoch": 17.465441470235717,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0101,
      "step": 43716
    },
    {
      "epoch": 17.465840990811028,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 43717
    },
    {
      "epoch": 17.466240511386335,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 43718
    },
    {
      "epoch": 17.466640031961646,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 43719
    },
    {
      "epoch": 17.467039552536956,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 43720
    },
    {
      "epoch": 17.467439073112267,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 43721
    },
    {
      "epoch": 17.467838593687574,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 43722
    },
    {
      "epoch": 17.468238114262885,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 43723
    },
    {
      "epoch": 17.468637634838196,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 43724
    },
    {
      "epoch": 17.469037155413503,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 43725
    },
    {
      "epoch": 17.469436675988813,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 43726
    },
    {
      "epoch": 17.469836196564124,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 43727
    },
    {
      "epoch": 17.47023571713943,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 43728
    },
    {
      "epoch": 17.470635237714742,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 43729
    },
    {
      "epoch": 17.471034758290052,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 43730
    },
    {
      "epoch": 17.471434278865363,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 43731
    },
    {
      "epoch": 17.47183379944067,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 43732
    },
    {
      "epoch": 17.47223332001598,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 43733
    },
    {
      "epoch": 17.47263284059129,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 43734
    },
    {
      "epoch": 17.4730323611666,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 43735
    },
    {
      "epoch": 17.47343188174191,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 43736
    },
    {
      "epoch": 17.47383140231722,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 43737
    },
    {
      "epoch": 17.474230922892527,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0116,
      "step": 43738
    },
    {
      "epoch": 17.474630443467838,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 43739
    },
    {
      "epoch": 17.47502996404315,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 43740
    },
    {
      "epoch": 17.47542948461846,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 43741
    },
    {
      "epoch": 17.475829005193766,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 43742
    },
    {
      "epoch": 17.476228525769077,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 43743
    },
    {
      "epoch": 17.476628046344388,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 43744
    },
    {
      "epoch": 17.477027566919695,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 43745
    },
    {
      "epoch": 17.477427087495006,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 43746
    },
    {
      "epoch": 17.477826608070316,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 43747
    },
    {
      "epoch": 17.478226128645627,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 43748
    },
    {
      "epoch": 17.478625649220934,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 43749
    },
    {
      "epoch": 17.479025169796245,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 43750
    },
    {
      "epoch": 17.479424690371555,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 43751
    },
    {
      "epoch": 17.479824210946862,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 43752
    },
    {
      "epoch": 17.480223731522173,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 43753
    },
    {
      "epoch": 17.480623252097484,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 43754
    },
    {
      "epoch": 17.48102277267279,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 43755
    },
    {
      "epoch": 17.4814222932481,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 43756
    },
    {
      "epoch": 17.481821813823412,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 43757
    },
    {
      "epoch": 17.482221334398723,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0125,
      "step": 43758
    },
    {
      "epoch": 17.48262085497403,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 43759
    },
    {
      "epoch": 17.48302037554934,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 43760
    },
    {
      "epoch": 17.48341989612465,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 43761
    },
    {
      "epoch": 17.48381941669996,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 43762
    },
    {
      "epoch": 17.48421893727527,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 43763
    },
    {
      "epoch": 17.48461845785058,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 43764
    },
    {
      "epoch": 17.48501797842589,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 43765
    },
    {
      "epoch": 17.485417499001198,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 43766
    },
    {
      "epoch": 17.48581701957651,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 43767
    },
    {
      "epoch": 17.48621654015182,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 43768
    },
    {
      "epoch": 17.486616060727126,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 43769
    },
    {
      "epoch": 17.487015581302437,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 43770
    },
    {
      "epoch": 17.487415101877748,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 43771
    },
    {
      "epoch": 17.487814622453055,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 43772
    },
    {
      "epoch": 17.488214143028365,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 43773
    },
    {
      "epoch": 17.488613663603676,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 43774
    },
    {
      "epoch": 17.489013184178987,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 43775
    },
    {
      "epoch": 17.489412704754294,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 43776
    },
    {
      "epoch": 17.489812225329604,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 43777
    },
    {
      "epoch": 17.490211745904915,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 43778
    },
    {
      "epoch": 17.490611266480222,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 43779
    },
    {
      "epoch": 17.491010787055533,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 43780
    },
    {
      "epoch": 17.491410307630844,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 43781
    },
    {
      "epoch": 17.491809828206154,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 43782
    },
    {
      "epoch": 17.49220934878146,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 43783
    },
    {
      "epoch": 17.492608869356772,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0094,
      "step": 43784
    },
    {
      "epoch": 17.493008389932083,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 43785
    },
    {
      "epoch": 17.49340791050739,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 43786
    },
    {
      "epoch": 17.4938074310827,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 43787
    },
    {
      "epoch": 17.49420695165801,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 43788
    },
    {
      "epoch": 17.49460647223332,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 43789
    },
    {
      "epoch": 17.49500599280863,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 43790
    },
    {
      "epoch": 17.49540551338394,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 43791
    },
    {
      "epoch": 17.49580503395925,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 43792
    },
    {
      "epoch": 17.496204554534557,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 43793
    },
    {
      "epoch": 17.496604075109868,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 43794
    },
    {
      "epoch": 17.49700359568518,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 43795
    },
    {
      "epoch": 17.497403116260486,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.011,
      "step": 43796
    },
    {
      "epoch": 17.497802636835797,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 43797
    },
    {
      "epoch": 17.498202157411107,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 43798
    },
    {
      "epoch": 17.498601677986418,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 43799
    },
    {
      "epoch": 17.499001198561725,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 43800
    },
    {
      "epoch": 17.499400719137036,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 43801
    },
    {
      "epoch": 17.499800239712346,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 43802
    },
    {
      "epoch": 17.500199760287654,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0095,
      "step": 43803
    },
    {
      "epoch": 17.500599280862964,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 43804
    },
    {
      "epoch": 17.500998801438275,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 43805
    },
    {
      "epoch": 17.501398322013582,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0122,
      "step": 43806
    },
    {
      "epoch": 17.501797842588893,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 43807
    },
    {
      "epoch": 17.502197363164203,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 43808
    },
    {
      "epoch": 17.502596883739514,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 43809
    },
    {
      "epoch": 17.50299640431482,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 43810
    },
    {
      "epoch": 17.503395924890132,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 43811
    },
    {
      "epoch": 17.503795445465443,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 43812
    },
    {
      "epoch": 17.50419496604075,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 43813
    },
    {
      "epoch": 17.50459448661606,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 43814
    },
    {
      "epoch": 17.50499400719137,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 43815
    },
    {
      "epoch": 17.50539352776668,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 43816
    },
    {
      "epoch": 17.50579304834199,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 43817
    },
    {
      "epoch": 17.5061925689173,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 43818
    },
    {
      "epoch": 17.50659208949261,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 43819
    },
    {
      "epoch": 17.506991610067917,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 43820
    },
    {
      "epoch": 17.507391130643228,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 43821
    },
    {
      "epoch": 17.50779065121854,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 43822
    },
    {
      "epoch": 17.508190171793846,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 43823
    },
    {
      "epoch": 17.508589692369156,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 43824
    },
    {
      "epoch": 17.508989212944467,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 43825
    },
    {
      "epoch": 17.509388733519778,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 43826
    },
    {
      "epoch": 17.509788254095085,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 43827
    },
    {
      "epoch": 17.510187774670396,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 43828
    },
    {
      "epoch": 17.510587295245706,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 43829
    },
    {
      "epoch": 17.510986815821013,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 43830
    },
    {
      "epoch": 17.511386336396324,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 43831
    },
    {
      "epoch": 17.511785856971635,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 43832
    },
    {
      "epoch": 17.512185377546945,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 43833
    },
    {
      "epoch": 17.512584898122252,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0104,
      "step": 43834
    },
    {
      "epoch": 17.512984418697563,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 43835
    },
    {
      "epoch": 17.513383939272874,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 43836
    },
    {
      "epoch": 17.51378345984818,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 43837
    },
    {
      "epoch": 17.51418298042349,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 43838
    },
    {
      "epoch": 17.514582500998802,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 43839
    },
    {
      "epoch": 17.51498202157411,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 43840
    },
    {
      "epoch": 17.51538154214942,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 43841
    },
    {
      "epoch": 17.51578106272473,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 43842
    },
    {
      "epoch": 17.51618058330004,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 43843
    },
    {
      "epoch": 17.51658010387535,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0084,
      "step": 43844
    },
    {
      "epoch": 17.51697962445066,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 43845
    },
    {
      "epoch": 17.51737914502597,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 43846
    },
    {
      "epoch": 17.517778665601277,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 43847
    },
    {
      "epoch": 17.518178186176588,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 43848
    },
    {
      "epoch": 17.5185777067519,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 43849
    },
    {
      "epoch": 17.518977227327206,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 43850
    },
    {
      "epoch": 17.519376747902516,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 43851
    },
    {
      "epoch": 17.519776268477827,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 43852
    },
    {
      "epoch": 17.520175789053138,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 43853
    },
    {
      "epoch": 17.520575309628445,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 43854
    },
    {
      "epoch": 17.520974830203755,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 43855
    },
    {
      "epoch": 17.521374350779066,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0107,
      "step": 43856
    },
    {
      "epoch": 17.521773871354373,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 43857
    },
    {
      "epoch": 17.522173391929684,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 43858
    },
    {
      "epoch": 17.522572912504994,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 43859
    },
    {
      "epoch": 17.522972433080305,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 43860
    },
    {
      "epoch": 17.523371953655612,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 43861
    },
    {
      "epoch": 17.523771474230923,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 43862
    },
    {
      "epoch": 17.524170994806234,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 43863
    },
    {
      "epoch": 17.52457051538154,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 43864
    },
    {
      "epoch": 17.52497003595685,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 43865
    },
    {
      "epoch": 17.525369556532162,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 43866
    },
    {
      "epoch": 17.52576907710747,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 43867
    },
    {
      "epoch": 17.52616859768278,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 43868
    },
    {
      "epoch": 17.52656811825809,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 43869
    },
    {
      "epoch": 17.5269676388334,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 43870
    },
    {
      "epoch": 17.52736715940871,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 43871
    },
    {
      "epoch": 17.52776667998402,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 43872
    },
    {
      "epoch": 17.52816620055933,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 43873
    },
    {
      "epoch": 17.528565721134637,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 43874
    },
    {
      "epoch": 17.528965241709948,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 43875
    },
    {
      "epoch": 17.529364762285258,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 43876
    },
    {
      "epoch": 17.52976428286057,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 43877
    },
    {
      "epoch": 17.530163803435876,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 43878
    },
    {
      "epoch": 17.530563324011187,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 43879
    },
    {
      "epoch": 17.530962844586497,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 43880
    },
    {
      "epoch": 17.531362365161804,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 43881
    },
    {
      "epoch": 17.531761885737115,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 43882
    },
    {
      "epoch": 17.532161406312426,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 43883
    },
    {
      "epoch": 17.532560926887733,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 43884
    },
    {
      "epoch": 17.532960447463044,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 43885
    },
    {
      "epoch": 17.533359968038354,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 43886
    },
    {
      "epoch": 17.533759488613665,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 43887
    },
    {
      "epoch": 17.534159009188972,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 43888
    },
    {
      "epoch": 17.534558529764283,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 43889
    },
    {
      "epoch": 17.534958050339593,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0118,
      "step": 43890
    },
    {
      "epoch": 17.5353575709149,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 43891
    },
    {
      "epoch": 17.53575709149021,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 43892
    },
    {
      "epoch": 17.536156612065522,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 43893
    },
    {
      "epoch": 17.536556132640833,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 43894
    },
    {
      "epoch": 17.53695565321614,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 43895
    },
    {
      "epoch": 17.53735517379145,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 43896
    },
    {
      "epoch": 17.53775469436676,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 43897
    },
    {
      "epoch": 17.538154214942068,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 43898
    },
    {
      "epoch": 17.53855373551738,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 43899
    },
    {
      "epoch": 17.53895325609269,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 43900
    },
    {
      "epoch": 17.539352776667997,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 43901
    },
    {
      "epoch": 17.539752297243307,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 43902
    },
    {
      "epoch": 17.540151817818618,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 43903
    },
    {
      "epoch": 17.54055133839393,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 43904
    },
    {
      "epoch": 17.540950858969236,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 43905
    },
    {
      "epoch": 17.541350379544546,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 43906
    },
    {
      "epoch": 17.541749900119857,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 43907
    },
    {
      "epoch": 17.542149420695164,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 43908
    },
    {
      "epoch": 17.542548941270475,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 43909
    },
    {
      "epoch": 17.542948461845786,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 43910
    },
    {
      "epoch": 17.543347982421096,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 43911
    },
    {
      "epoch": 17.543747502996403,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 43912
    },
    {
      "epoch": 17.544147023571714,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 43913
    },
    {
      "epoch": 17.544546544147025,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 43914
    },
    {
      "epoch": 17.544946064722332,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 43915
    },
    {
      "epoch": 17.545345585297643,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 43916
    },
    {
      "epoch": 17.545745105872953,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 43917
    },
    {
      "epoch": 17.54614462644826,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 43918
    },
    {
      "epoch": 17.54654414702357,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 43919
    },
    {
      "epoch": 17.54694366759888,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 43920
    },
    {
      "epoch": 17.547343188174192,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 43921
    },
    {
      "epoch": 17.5477427087495,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 43922
    },
    {
      "epoch": 17.54814222932481,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 43923
    },
    {
      "epoch": 17.54854174990012,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 43924
    },
    {
      "epoch": 17.548941270475428,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 43925
    },
    {
      "epoch": 17.54934079105074,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 43926
    },
    {
      "epoch": 17.54974031162605,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 43927
    },
    {
      "epoch": 17.55013983220136,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 43928
    },
    {
      "epoch": 17.550539352776667,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 43929
    },
    {
      "epoch": 17.550938873351978,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 43930
    },
    {
      "epoch": 17.55133839392729,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 43931
    },
    {
      "epoch": 17.551737914502596,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 43932
    },
    {
      "epoch": 17.552137435077906,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 43933
    },
    {
      "epoch": 17.552536955653217,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 43934
    },
    {
      "epoch": 17.552936476228524,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 43935
    },
    {
      "epoch": 17.553335996803835,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 43936
    },
    {
      "epoch": 17.553735517379145,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 43937
    },
    {
      "epoch": 17.554135037954456,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 43938
    },
    {
      "epoch": 17.554534558529763,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 43939
    },
    {
      "epoch": 17.554934079105074,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 43940
    },
    {
      "epoch": 17.555333599680385,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 43941
    },
    {
      "epoch": 17.55573312025569,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 43942
    },
    {
      "epoch": 17.556132640831002,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 43943
    },
    {
      "epoch": 17.556532161406313,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0126,
      "step": 43944
    },
    {
      "epoch": 17.556931681981624,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 43945
    },
    {
      "epoch": 17.55733120255693,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 43946
    },
    {
      "epoch": 17.55773072313224,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 43947
    },
    {
      "epoch": 17.558130243707552,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 43948
    },
    {
      "epoch": 17.55852976428286,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 43949
    },
    {
      "epoch": 17.55892928485817,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 43950
    },
    {
      "epoch": 17.55932880543348,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0115,
      "step": 43951
    },
    {
      "epoch": 17.559728326008788,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 43952
    },
    {
      "epoch": 17.5601278465841,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 43953
    },
    {
      "epoch": 17.56052736715941,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 43954
    },
    {
      "epoch": 17.56092688773472,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 43955
    },
    {
      "epoch": 17.561326408310027,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 43956
    },
    {
      "epoch": 17.561725928885338,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 43957
    },
    {
      "epoch": 17.56212544946065,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 43958
    },
    {
      "epoch": 17.562524970035955,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 43959
    },
    {
      "epoch": 17.562924490611266,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 43960
    },
    {
      "epoch": 17.563324011186577,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 43961
    },
    {
      "epoch": 17.563723531761887,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 43962
    },
    {
      "epoch": 17.564123052337195,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 43963
    },
    {
      "epoch": 17.564522572912505,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 43964
    },
    {
      "epoch": 17.564922093487816,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 43965
    },
    {
      "epoch": 17.565321614063123,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 43966
    },
    {
      "epoch": 17.565721134638434,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 43967
    },
    {
      "epoch": 17.566120655213744,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 43968
    },
    {
      "epoch": 17.56652017578905,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 43969
    },
    {
      "epoch": 17.566919696364362,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 43970
    },
    {
      "epoch": 17.567319216939673,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 43971
    },
    {
      "epoch": 17.567718737514983,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 43972
    },
    {
      "epoch": 17.56811825809029,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 43973
    },
    {
      "epoch": 17.5685177786656,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 43974
    },
    {
      "epoch": 17.568917299240912,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 43975
    },
    {
      "epoch": 17.56931681981622,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 43976
    },
    {
      "epoch": 17.56971634039153,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 43977
    },
    {
      "epoch": 17.57011586096684,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 43978
    },
    {
      "epoch": 17.57051538154215,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 43979
    },
    {
      "epoch": 17.570914902117458,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 43980
    },
    {
      "epoch": 17.57131442269277,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 43981
    },
    {
      "epoch": 17.57171394326808,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 43982
    },
    {
      "epoch": 17.572113463843387,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 43983
    },
    {
      "epoch": 17.572512984418697,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 43984
    },
    {
      "epoch": 17.572912504994008,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 43985
    },
    {
      "epoch": 17.573312025569315,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 43986
    },
    {
      "epoch": 17.573711546144626,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 43987
    },
    {
      "epoch": 17.574111066719937,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 43988
    },
    {
      "epoch": 17.574510587295247,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 43989
    },
    {
      "epoch": 17.574910107870554,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 43990
    },
    {
      "epoch": 17.575309628445865,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 43991
    },
    {
      "epoch": 17.575709149021176,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 43992
    },
    {
      "epoch": 17.576108669596483,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 43993
    },
    {
      "epoch": 17.576508190171793,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0111,
      "step": 43994
    },
    {
      "epoch": 17.576907710747104,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 43995
    },
    {
      "epoch": 17.577307231322415,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 43996
    },
    {
      "epoch": 17.577706751897722,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 43997
    },
    {
      "epoch": 17.578106272473033,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 43998
    },
    {
      "epoch": 17.578505793048343,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 43999
    },
    {
      "epoch": 17.57890531362365,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 44000
    },
    {
      "epoch": 17.57930483419896,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 44001
    },
    {
      "epoch": 17.57970435477427,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 44002
    },
    {
      "epoch": 17.58010387534958,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 44003
    },
    {
      "epoch": 17.58050339592489,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 44004
    },
    {
      "epoch": 17.5809029165002,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 44005
    },
    {
      "epoch": 17.58130243707551,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 44006
    },
    {
      "epoch": 17.581701957650818,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 44007
    },
    {
      "epoch": 17.58210147822613,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 44008
    },
    {
      "epoch": 17.58250099880144,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 44009
    },
    {
      "epoch": 17.582900519376746,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 44010
    },
    {
      "epoch": 17.583300039952057,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 44011
    },
    {
      "epoch": 17.583699560527368,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 44012
    },
    {
      "epoch": 17.58409908110268,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 44013
    },
    {
      "epoch": 17.584498601677986,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 44014
    },
    {
      "epoch": 17.584898122253296,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 44015
    },
    {
      "epoch": 17.585297642828607,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 44016
    },
    {
      "epoch": 17.585697163403914,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 44017
    },
    {
      "epoch": 17.586096683979225,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 44018
    },
    {
      "epoch": 17.586496204554535,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 44019
    },
    {
      "epoch": 17.586895725129843,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 44020
    },
    {
      "epoch": 17.587295245705153,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 44021
    },
    {
      "epoch": 17.587694766280464,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 44022
    },
    {
      "epoch": 17.588094286855775,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 44023
    },
    {
      "epoch": 17.58849380743108,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 44024
    },
    {
      "epoch": 17.588893328006392,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 44025
    },
    {
      "epoch": 17.589292848581703,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 44026
    },
    {
      "epoch": 17.58969236915701,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 44027
    },
    {
      "epoch": 17.59009188973232,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 44028
    },
    {
      "epoch": 17.59049141030763,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 44029
    },
    {
      "epoch": 17.590890930882942,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 44030
    },
    {
      "epoch": 17.59129045145825,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 44031
    },
    {
      "epoch": 17.59168997203356,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 44032
    },
    {
      "epoch": 17.59208949260887,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 44033
    },
    {
      "epoch": 17.592489013184178,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 44034
    },
    {
      "epoch": 17.59288853375949,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 44035
    },
    {
      "epoch": 17.5932880543348,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 44036
    },
    {
      "epoch": 17.593687574910106,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0074,
      "step": 44037
    },
    {
      "epoch": 17.594087095485417,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 44038
    },
    {
      "epoch": 17.594486616060728,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 44039
    },
    {
      "epoch": 17.59488613663604,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 44040
    },
    {
      "epoch": 17.595285657211345,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 44041
    },
    {
      "epoch": 17.595685177786656,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 44042
    },
    {
      "epoch": 17.596084698361967,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 44043
    },
    {
      "epoch": 17.596484218937274,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 44044
    },
    {
      "epoch": 17.596883739512585,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 44045
    },
    {
      "epoch": 17.597283260087895,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 44046
    },
    {
      "epoch": 17.597682780663206,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 44047
    },
    {
      "epoch": 17.598082301238513,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 44048
    },
    {
      "epoch": 17.598481821813824,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 44049
    },
    {
      "epoch": 17.598881342389134,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 44050
    },
    {
      "epoch": 17.59928086296444,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 44051
    },
    {
      "epoch": 17.599680383539752,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 44052
    },
    {
      "epoch": 17.600079904115063,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 44053
    },
    {
      "epoch": 17.60047942469037,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 44054
    },
    {
      "epoch": 17.60087894526568,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 44055
    },
    {
      "epoch": 17.60127846584099,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 44056
    },
    {
      "epoch": 17.601677986416302,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 44057
    },
    {
      "epoch": 17.60207750699161,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 44058
    },
    {
      "epoch": 17.60247702756692,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 44059
    },
    {
      "epoch": 17.60287654814223,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 44060
    },
    {
      "epoch": 17.603276068717538,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 44061
    },
    {
      "epoch": 17.60367558929285,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 44062
    },
    {
      "epoch": 17.60407510986816,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 44063
    },
    {
      "epoch": 17.60447463044347,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 44064
    },
    {
      "epoch": 17.604874151018777,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 44065
    },
    {
      "epoch": 17.605273671594087,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 44066
    },
    {
      "epoch": 17.605673192169398,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 44067
    },
    {
      "epoch": 17.606072712744705,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 44068
    },
    {
      "epoch": 17.606472233320016,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 44069
    },
    {
      "epoch": 17.606871753895327,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 44070
    },
    {
      "epoch": 17.607271274470634,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 44071
    },
    {
      "epoch": 17.607670795045944,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 44072
    },
    {
      "epoch": 17.608070315621255,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 44073
    },
    {
      "epoch": 17.608469836196566,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 44074
    },
    {
      "epoch": 17.608869356771873,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 44075
    },
    {
      "epoch": 17.609268877347183,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 44076
    },
    {
      "epoch": 17.609668397922494,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 44077
    },
    {
      "epoch": 17.6100679184978,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 44078
    },
    {
      "epoch": 17.610467439073112,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 44079
    },
    {
      "epoch": 17.610866959648423,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 44080
    },
    {
      "epoch": 17.611266480223733,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 44081
    },
    {
      "epoch": 17.61166600079904,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 44082
    },
    {
      "epoch": 17.61206552137435,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.038,
      "step": 44083
    },
    {
      "epoch": 17.612465041949662,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 44084
    },
    {
      "epoch": 17.61286456252497,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 44085
    },
    {
      "epoch": 17.61326408310028,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 44086
    },
    {
      "epoch": 17.61366360367559,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 44087
    },
    {
      "epoch": 17.614063124250897,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 44088
    },
    {
      "epoch": 17.614462644826208,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 44089
    },
    {
      "epoch": 17.61486216540152,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 44090
    },
    {
      "epoch": 17.61526168597683,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0118,
      "step": 44091
    },
    {
      "epoch": 17.615661206552137,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 44092
    },
    {
      "epoch": 17.616060727127447,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 44093
    },
    {
      "epoch": 17.616460247702758,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 44094
    },
    {
      "epoch": 17.616859768278065,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 44095
    },
    {
      "epoch": 17.617259288853376,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 44096
    },
    {
      "epoch": 17.617658809428686,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 44097
    },
    {
      "epoch": 17.618058330003997,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 44098
    },
    {
      "epoch": 17.618457850579304,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 44099
    },
    {
      "epoch": 17.618857371154615,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 44100
    },
    {
      "epoch": 17.619256891729925,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 44101
    },
    {
      "epoch": 17.619656412305233,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 44102
    },
    {
      "epoch": 17.620055932880543,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 44103
    },
    {
      "epoch": 17.620455453455854,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 44104
    },
    {
      "epoch": 17.62085497403116,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 44105
    },
    {
      "epoch": 17.62125449460647,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 44106
    },
    {
      "epoch": 17.621654015181782,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 44107
    },
    {
      "epoch": 17.622053535757093,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 44108
    },
    {
      "epoch": 17.6224530563324,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 44109
    },
    {
      "epoch": 17.62285257690771,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 44110
    },
    {
      "epoch": 17.62325209748302,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 44111
    },
    {
      "epoch": 17.62365161805833,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 44112
    },
    {
      "epoch": 17.62405113863364,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 44113
    },
    {
      "epoch": 17.62445065920895,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 44114
    },
    {
      "epoch": 17.62485017978426,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 44115
    },
    {
      "epoch": 17.625249700359568,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 44116
    },
    {
      "epoch": 17.62564922093488,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 44117
    },
    {
      "epoch": 17.62604874151019,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 44118
    },
    {
      "epoch": 17.626448262085496,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 44119
    },
    {
      "epoch": 17.626847782660807,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 44120
    },
    {
      "epoch": 17.627247303236118,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 44121
    },
    {
      "epoch": 17.627646823811425,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 44122
    },
    {
      "epoch": 17.628046344386735,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 44123
    },
    {
      "epoch": 17.628445864962046,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 44124
    },
    {
      "epoch": 17.628845385537357,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 44125
    },
    {
      "epoch": 17.629244906112664,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 44126
    },
    {
      "epoch": 17.629644426687975,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 44127
    },
    {
      "epoch": 17.630043947263285,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 44128
    },
    {
      "epoch": 17.630443467838592,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 44129
    },
    {
      "epoch": 17.630842988413903,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 44130
    },
    {
      "epoch": 17.631242508989214,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 44131
    },
    {
      "epoch": 17.63164202956452,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 44132
    },
    {
      "epoch": 17.63204155013983,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 44133
    },
    {
      "epoch": 17.632441070715142,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 44134
    },
    {
      "epoch": 17.632840591290453,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 44135
    },
    {
      "epoch": 17.63324011186576,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 44136
    },
    {
      "epoch": 17.63363963244107,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 44137
    },
    {
      "epoch": 17.63403915301638,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0113,
      "step": 44138
    },
    {
      "epoch": 17.63443867359169,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 44139
    },
    {
      "epoch": 17.634838194167,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 44140
    },
    {
      "epoch": 17.63523771474231,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 44141
    },
    {
      "epoch": 17.63563723531762,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 44142
    },
    {
      "epoch": 17.636036755892928,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 44143
    },
    {
      "epoch": 17.63643627646824,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 44144
    },
    {
      "epoch": 17.63683579704355,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0447,
      "step": 44145
    },
    {
      "epoch": 17.637235317618856,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 44146
    },
    {
      "epoch": 17.637634838194167,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0384,
      "step": 44147
    },
    {
      "epoch": 17.638034358769477,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 44148
    },
    {
      "epoch": 17.638433879344785,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 44149
    },
    {
      "epoch": 17.638833399920095,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 44150
    },
    {
      "epoch": 17.639232920495406,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 44151
    },
    {
      "epoch": 17.639632441070717,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 44152
    },
    {
      "epoch": 17.640031961646024,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0108,
      "step": 44153
    },
    {
      "epoch": 17.640431482221334,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 44154
    },
    {
      "epoch": 17.640831002796645,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 44155
    },
    {
      "epoch": 17.641230523371952,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 44156
    },
    {
      "epoch": 17.641630043947263,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 44157
    },
    {
      "epoch": 17.642029564522574,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0118,
      "step": 44158
    },
    {
      "epoch": 17.642429085097884,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 44159
    },
    {
      "epoch": 17.64282860567319,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 44160
    },
    {
      "epoch": 17.643228126248502,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 44161
    },
    {
      "epoch": 17.643627646823813,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 44162
    },
    {
      "epoch": 17.64402716739912,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 44163
    },
    {
      "epoch": 17.64442668797443,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 44164
    },
    {
      "epoch": 17.64482620854974,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 44165
    },
    {
      "epoch": 17.64522572912505,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 44166
    },
    {
      "epoch": 17.64562524970036,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 44167
    },
    {
      "epoch": 17.64602477027567,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 44168
    },
    {
      "epoch": 17.64642429085098,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 44169
    },
    {
      "epoch": 17.646823811426287,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 44170
    },
    {
      "epoch": 17.647223332001598,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 44171
    },
    {
      "epoch": 17.64762285257691,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.009,
      "step": 44172
    },
    {
      "epoch": 17.648022373152216,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 44173
    },
    {
      "epoch": 17.648421893727527,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 44174
    },
    {
      "epoch": 17.648821414302837,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 44175
    },
    {
      "epoch": 17.649220934878148,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 44176
    },
    {
      "epoch": 17.649620455453455,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 44177
    },
    {
      "epoch": 17.650019976028766,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 44178
    },
    {
      "epoch": 17.650419496604076,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 44179
    },
    {
      "epoch": 17.650819017179384,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 44180
    },
    {
      "epoch": 17.651218537754694,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 44181
    },
    {
      "epoch": 17.651618058330005,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0441,
      "step": 44182
    },
    {
      "epoch": 17.652017578905312,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 44183
    },
    {
      "epoch": 17.652417099480623,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 44184
    },
    {
      "epoch": 17.652816620055933,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 44185
    },
    {
      "epoch": 17.653216140631244,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 44186
    },
    {
      "epoch": 17.65361566120655,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 44187
    },
    {
      "epoch": 17.654015181781862,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 44188
    },
    {
      "epoch": 17.654414702357172,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0088,
      "step": 44189
    },
    {
      "epoch": 17.65481422293248,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 44190
    },
    {
      "epoch": 17.65521374350779,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 44191
    },
    {
      "epoch": 17.6556132640831,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0104,
      "step": 44192
    },
    {
      "epoch": 17.65601278465841,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 44193
    },
    {
      "epoch": 17.65641230523372,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 44194
    },
    {
      "epoch": 17.65681182580903,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 44195
    },
    {
      "epoch": 17.65721134638434,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 44196
    },
    {
      "epoch": 17.657610866959647,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 44197
    },
    {
      "epoch": 17.658010387534958,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 44198
    },
    {
      "epoch": 17.65840990811027,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 44199
    },
    {
      "epoch": 17.658809428685576,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 44200
    },
    {
      "epoch": 17.659208949260886,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 44201
    },
    {
      "epoch": 17.659608469836197,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 44202
    },
    {
      "epoch": 17.660007990411508,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 44203
    },
    {
      "epoch": 17.660407510986815,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 44204
    },
    {
      "epoch": 17.660807031562126,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 44205
    },
    {
      "epoch": 17.661206552137436,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 44206
    },
    {
      "epoch": 17.661606072712743,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 44207
    },
    {
      "epoch": 17.662005593288054,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 44208
    },
    {
      "epoch": 17.662405113863365,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 44209
    },
    {
      "epoch": 17.66280463443867,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 44210
    },
    {
      "epoch": 17.663204155013982,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 44211
    },
    {
      "epoch": 17.663603675589293,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 44212
    },
    {
      "epoch": 17.664003196164604,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 44213
    },
    {
      "epoch": 17.66440271673991,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 44214
    },
    {
      "epoch": 17.66480223731522,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 44215
    },
    {
      "epoch": 17.665201757890532,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 44216
    },
    {
      "epoch": 17.66560127846584,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 44217
    },
    {
      "epoch": 17.66600079904115,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 44218
    },
    {
      "epoch": 17.66640031961646,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 44219
    },
    {
      "epoch": 17.66679984019177,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 44220
    },
    {
      "epoch": 17.66719936076708,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 44221
    },
    {
      "epoch": 17.66759888134239,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 44222
    },
    {
      "epoch": 17.6679984019177,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 44223
    },
    {
      "epoch": 17.668397922493007,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 44224
    },
    {
      "epoch": 17.668797443068318,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 44225
    },
    {
      "epoch": 17.66919696364363,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 44226
    },
    {
      "epoch": 17.669596484218935,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 44227
    },
    {
      "epoch": 17.669996004794246,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 44228
    },
    {
      "epoch": 17.670395525369557,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 44229
    },
    {
      "epoch": 17.670795045944867,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 44230
    },
    {
      "epoch": 17.671194566520175,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 44231
    },
    {
      "epoch": 17.671594087095485,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 44232
    },
    {
      "epoch": 17.671993607670796,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 44233
    },
    {
      "epoch": 17.672393128246103,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 44234
    },
    {
      "epoch": 17.672792648821414,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 44235
    },
    {
      "epoch": 17.673192169396724,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 44236
    },
    {
      "epoch": 17.673591689972035,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 44237
    },
    {
      "epoch": 17.673991210547342,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 44238
    },
    {
      "epoch": 17.674390731122653,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 44239
    },
    {
      "epoch": 17.674790251697964,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 44240
    },
    {
      "epoch": 17.67518977227327,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0075,
      "step": 44241
    },
    {
      "epoch": 17.67558929284858,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 44242
    },
    {
      "epoch": 17.675988813423892,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 44243
    },
    {
      "epoch": 17.6763883339992,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0402,
      "step": 44244
    },
    {
      "epoch": 17.67678785457451,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 44245
    },
    {
      "epoch": 17.67718737514982,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 44246
    },
    {
      "epoch": 17.67758689572513,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 44247
    },
    {
      "epoch": 17.67798641630044,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 44248
    },
    {
      "epoch": 17.67838593687575,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 44249
    },
    {
      "epoch": 17.67878545745106,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 44250
    },
    {
      "epoch": 17.679184978026367,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 44251
    },
    {
      "epoch": 17.679584498601677,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 44252
    },
    {
      "epoch": 17.679984019176988,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 44253
    },
    {
      "epoch": 17.6803835397523,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 44254
    },
    {
      "epoch": 17.680783060327606,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 44255
    },
    {
      "epoch": 17.681182580902917,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 44256
    },
    {
      "epoch": 17.681582101478227,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 44257
    },
    {
      "epoch": 17.681981622053534,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 44258
    },
    {
      "epoch": 17.682381142628845,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 44259
    },
    {
      "epoch": 17.682780663204156,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0117,
      "step": 44260
    },
    {
      "epoch": 17.683180183779463,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 44261
    },
    {
      "epoch": 17.683579704354774,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 44262
    },
    {
      "epoch": 17.683979224930084,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 44263
    },
    {
      "epoch": 17.684378745505395,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 44264
    },
    {
      "epoch": 17.684778266080702,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 44265
    },
    {
      "epoch": 17.685177786656013,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 44266
    },
    {
      "epoch": 17.685577307231323,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0118,
      "step": 44267
    },
    {
      "epoch": 17.68597682780663,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 44268
    },
    {
      "epoch": 17.68637634838194,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 44269
    },
    {
      "epoch": 17.686775868957252,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 44270
    },
    {
      "epoch": 17.687175389532563,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 44271
    },
    {
      "epoch": 17.68757491010787,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 44272
    },
    {
      "epoch": 17.68797443068318,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 44273
    },
    {
      "epoch": 17.68837395125849,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0116,
      "step": 44274
    },
    {
      "epoch": 17.688773471833798,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 44275
    },
    {
      "epoch": 17.68917299240911,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 44276
    },
    {
      "epoch": 17.68957251298442,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 44277
    },
    {
      "epoch": 17.689972033559727,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 44278
    },
    {
      "epoch": 17.690371554135037,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 44279
    },
    {
      "epoch": 17.690771074710348,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 44280
    },
    {
      "epoch": 17.69117059528566,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 44281
    },
    {
      "epoch": 17.691570115860966,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0126,
      "step": 44282
    },
    {
      "epoch": 17.691969636436276,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 44283
    },
    {
      "epoch": 17.692369157011587,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 44284
    },
    {
      "epoch": 17.692768677586894,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 44285
    },
    {
      "epoch": 17.693168198162205,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 44286
    },
    {
      "epoch": 17.693567718737516,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 44287
    },
    {
      "epoch": 17.693967239312826,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 44288
    },
    {
      "epoch": 17.694366759888133,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 44289
    },
    {
      "epoch": 17.694766280463444,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 44290
    },
    {
      "epoch": 17.695165801038755,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 44291
    },
    {
      "epoch": 17.695565321614062,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 44292
    },
    {
      "epoch": 17.695964842189372,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 44293
    },
    {
      "epoch": 17.696364362764683,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 44294
    },
    {
      "epoch": 17.69676388333999,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 44295
    },
    {
      "epoch": 17.6971634039153,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 44296
    },
    {
      "epoch": 17.69756292449061,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 44297
    },
    {
      "epoch": 17.697962445065922,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0412,
      "step": 44298
    },
    {
      "epoch": 17.69836196564123,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 44299
    },
    {
      "epoch": 17.69876148621654,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0428,
      "step": 44300
    },
    {
      "epoch": 17.69916100679185,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 44301
    },
    {
      "epoch": 17.699560527367158,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 44302
    },
    {
      "epoch": 17.69996004794247,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 44303
    },
    {
      "epoch": 17.70035956851778,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 44304
    },
    {
      "epoch": 17.70075908909309,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 44305
    },
    {
      "epoch": 17.701158609668397,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 44306
    },
    {
      "epoch": 17.701558130243708,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 44307
    },
    {
      "epoch": 17.70195765081902,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 44308
    },
    {
      "epoch": 17.702357171394326,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 44309
    },
    {
      "epoch": 17.702756691969636,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 44310
    },
    {
      "epoch": 17.703156212544947,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 44311
    },
    {
      "epoch": 17.703555733120254,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 44312
    },
    {
      "epoch": 17.703955253695565,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 44313
    },
    {
      "epoch": 17.704354774270875,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 44314
    },
    {
      "epoch": 17.704754294846186,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 44315
    },
    {
      "epoch": 17.705153815421493,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 44316
    },
    {
      "epoch": 17.705553335996804,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 44317
    },
    {
      "epoch": 17.705952856572114,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0122,
      "step": 44318
    },
    {
      "epoch": 17.70635237714742,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 44319
    },
    {
      "epoch": 17.706751897722732,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 44320
    },
    {
      "epoch": 17.707151418298043,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 44321
    },
    {
      "epoch": 17.707550938873354,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 44322
    },
    {
      "epoch": 17.70795045944866,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 44323
    },
    {
      "epoch": 17.70834998002397,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 44324
    },
    {
      "epoch": 17.708749500599282,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 44325
    },
    {
      "epoch": 17.70914902117459,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 44326
    },
    {
      "epoch": 17.7095485417499,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 44327
    },
    {
      "epoch": 17.70994806232521,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 44328
    },
    {
      "epoch": 17.710347582900518,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 44329
    },
    {
      "epoch": 17.71074710347583,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 44330
    },
    {
      "epoch": 17.71114662405114,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 44331
    },
    {
      "epoch": 17.71154614462645,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 44332
    },
    {
      "epoch": 17.711945665201757,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 44333
    },
    {
      "epoch": 17.712345185777068,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 44334
    },
    {
      "epoch": 17.712744706352378,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 44335
    },
    {
      "epoch": 17.713144226927685,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0415,
      "step": 44336
    },
    {
      "epoch": 17.713543747502996,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 44337
    },
    {
      "epoch": 17.713943268078307,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0043,
      "step": 44338
    },
    {
      "epoch": 17.714342788653617,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 44339
    },
    {
      "epoch": 17.714742309228924,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 44340
    },
    {
      "epoch": 17.715141829804235,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 44341
    },
    {
      "epoch": 17.715541350379546,
      "grad_norm": 0.036865234375,
      "learning_rate": 0.0005,
      "loss": 1.0091,
      "step": 44342
    },
    {
      "epoch": 17.715940870954853,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 44343
    },
    {
      "epoch": 17.716340391530164,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 44344
    },
    {
      "epoch": 17.716739912105474,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 44345
    },
    {
      "epoch": 17.71713943268078,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 44346
    },
    {
      "epoch": 17.717538953256092,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 44347
    },
    {
      "epoch": 17.717938473831403,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 44348
    },
    {
      "epoch": 17.718337994406713,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 44349
    },
    {
      "epoch": 17.71873751498202,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 44350
    },
    {
      "epoch": 17.71913703555733,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 44351
    },
    {
      "epoch": 17.719536556132642,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 44352
    },
    {
      "epoch": 17.71993607670795,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 44353
    },
    {
      "epoch": 17.72033559728326,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 44354
    },
    {
      "epoch": 17.72073511785857,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 44355
    },
    {
      "epoch": 17.72113463843388,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 44356
    },
    {
      "epoch": 17.721534159009188,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 44357
    },
    {
      "epoch": 17.7219336795845,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 44358
    },
    {
      "epoch": 17.72233320015981,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 44359
    },
    {
      "epoch": 17.722732720735117,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 44360
    },
    {
      "epoch": 17.723132241310427,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 44361
    },
    {
      "epoch": 17.723531761885738,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 44362
    },
    {
      "epoch": 17.723931282461045,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 44363
    },
    {
      "epoch": 17.724330803036356,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0097,
      "step": 44364
    },
    {
      "epoch": 17.724730323611666,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 44365
    },
    {
      "epoch": 17.725129844186977,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 44366
    },
    {
      "epoch": 17.725529364762284,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 44367
    },
    {
      "epoch": 17.725928885337595,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 44368
    },
    {
      "epoch": 17.726328405912906,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 44369
    },
    {
      "epoch": 17.726727926488213,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 44370
    },
    {
      "epoch": 17.727127447063523,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 44371
    },
    {
      "epoch": 17.727526967638834,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 44372
    },
    {
      "epoch": 17.727926488214145,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 44373
    },
    {
      "epoch": 17.728326008789452,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0054,
      "step": 44374
    },
    {
      "epoch": 17.728725529364763,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 44375
    },
    {
      "epoch": 17.729125049940073,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 44376
    },
    {
      "epoch": 17.72952457051538,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 44377
    },
    {
      "epoch": 17.72992409109069,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 44378
    },
    {
      "epoch": 17.730323611666,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 44379
    },
    {
      "epoch": 17.73072313224131,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 44380
    },
    {
      "epoch": 17.73112265281662,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 44381
    },
    {
      "epoch": 17.73152217339193,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 44382
    },
    {
      "epoch": 17.73192169396724,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0127,
      "step": 44383
    },
    {
      "epoch": 17.732321214542548,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 44384
    },
    {
      "epoch": 17.73272073511786,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 44385
    },
    {
      "epoch": 17.73312025569317,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 44386
    },
    {
      "epoch": 17.733519776268476,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 44387
    },
    {
      "epoch": 17.733919296843787,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 44388
    },
    {
      "epoch": 17.734318817419098,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 44389
    },
    {
      "epoch": 17.73471833799441,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 44390
    },
    {
      "epoch": 17.735117858569716,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 44391
    },
    {
      "epoch": 17.735517379145026,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 44392
    },
    {
      "epoch": 17.735916899720337,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 44393
    },
    {
      "epoch": 17.736316420295644,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 44394
    },
    {
      "epoch": 17.736715940870955,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.009,
      "step": 44395
    },
    {
      "epoch": 17.737115461446265,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0119,
      "step": 44396
    },
    {
      "epoch": 17.737514982021573,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 44397
    },
    {
      "epoch": 17.737914502596883,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 44398
    },
    {
      "epoch": 17.738314023172194,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 44399
    },
    {
      "epoch": 17.738713543747505,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 44400
    },
    {
      "epoch": 17.73911306432281,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 44401
    },
    {
      "epoch": 17.739512584898122,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 44402
    },
    {
      "epoch": 17.739912105473433,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 44403
    },
    {
      "epoch": 17.74031162604874,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 44404
    },
    {
      "epoch": 17.74071114662405,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 44405
    },
    {
      "epoch": 17.74111066719936,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 44406
    },
    {
      "epoch": 17.741510187774672,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 44407
    },
    {
      "epoch": 17.74190970834998,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 44408
    },
    {
      "epoch": 17.74230922892529,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 44409
    },
    {
      "epoch": 17.7427087495006,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 44410
    },
    {
      "epoch": 17.743108270075908,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 44411
    },
    {
      "epoch": 17.74350779065122,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 44412
    },
    {
      "epoch": 17.74390731122653,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 44413
    },
    {
      "epoch": 17.744306831801836,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 44414
    },
    {
      "epoch": 17.744706352377147,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 44415
    },
    {
      "epoch": 17.745105872952458,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 44416
    },
    {
      "epoch": 17.74550539352777,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 44417
    },
    {
      "epoch": 17.745904914103075,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 44418
    },
    {
      "epoch": 17.746304434678386,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 44419
    },
    {
      "epoch": 17.746703955253697,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 44420
    },
    {
      "epoch": 17.747103475829004,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 44421
    },
    {
      "epoch": 17.747502996404315,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 44422
    },
    {
      "epoch": 17.747902516979625,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 44423
    },
    {
      "epoch": 17.748302037554936,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 44424
    },
    {
      "epoch": 17.748701558130243,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 44425
    },
    {
      "epoch": 17.749101078705554,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 44426
    },
    {
      "epoch": 17.749500599280864,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 44427
    },
    {
      "epoch": 17.74990011985617,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 44428
    },
    {
      "epoch": 17.750299640431482,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 44429
    },
    {
      "epoch": 17.750699161006793,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 44430
    },
    {
      "epoch": 17.7510986815821,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 44431
    },
    {
      "epoch": 17.75149820215741,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 44432
    },
    {
      "epoch": 17.75189772273272,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 44433
    },
    {
      "epoch": 17.752297243308032,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 44434
    },
    {
      "epoch": 17.75269676388334,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 44435
    },
    {
      "epoch": 17.75309628445865,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 44436
    },
    {
      "epoch": 17.75349580503396,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 44437
    },
    {
      "epoch": 17.753895325609268,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 44438
    },
    {
      "epoch": 17.754294846184578,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 44439
    },
    {
      "epoch": 17.75469436675989,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 44440
    },
    {
      "epoch": 17.7550938873352,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 44441
    },
    {
      "epoch": 17.755493407910507,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 44442
    },
    {
      "epoch": 17.755892928485817,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 44443
    },
    {
      "epoch": 17.756292449061128,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 44444
    },
    {
      "epoch": 17.756691969636435,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 44445
    },
    {
      "epoch": 17.757091490211746,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 44446
    },
    {
      "epoch": 17.757491010787056,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 44447
    },
    {
      "epoch": 17.757890531362364,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 44448
    },
    {
      "epoch": 17.758290051937674,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 44449
    },
    {
      "epoch": 17.758689572512985,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 44450
    },
    {
      "epoch": 17.759089093088296,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 44451
    },
    {
      "epoch": 17.759488613663603,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 44452
    },
    {
      "epoch": 17.759888134238913,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 44453
    },
    {
      "epoch": 17.760287654814224,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 44454
    },
    {
      "epoch": 17.76068717538953,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 44455
    },
    {
      "epoch": 17.761086695964842,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 44456
    },
    {
      "epoch": 17.761486216540153,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 44457
    },
    {
      "epoch": 17.761885737115463,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 44458
    },
    {
      "epoch": 17.76228525769077,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 44459
    },
    {
      "epoch": 17.76268477826608,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 44460
    },
    {
      "epoch": 17.76308429884139,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 44461
    },
    {
      "epoch": 17.7634838194167,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 44462
    },
    {
      "epoch": 17.76388333999201,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 44463
    },
    {
      "epoch": 17.76428286056732,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 44464
    },
    {
      "epoch": 17.764682381142627,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 44465
    },
    {
      "epoch": 17.765081901717938,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 44466
    },
    {
      "epoch": 17.76548142229325,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0102,
      "step": 44467
    },
    {
      "epoch": 17.76588094286856,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 44468
    },
    {
      "epoch": 17.766280463443866,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 44469
    },
    {
      "epoch": 17.766679984019177,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 44470
    },
    {
      "epoch": 17.767079504594488,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 44471
    },
    {
      "epoch": 17.767479025169795,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 44472
    },
    {
      "epoch": 17.767878545745106,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 44473
    },
    {
      "epoch": 17.768278066320416,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 44474
    },
    {
      "epoch": 17.768677586895727,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 44475
    },
    {
      "epoch": 17.769077107471034,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 44476
    },
    {
      "epoch": 17.769476628046345,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 44477
    },
    {
      "epoch": 17.769876148621655,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 44478
    },
    {
      "epoch": 17.770275669196963,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 44479
    },
    {
      "epoch": 17.770675189772273,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 44480
    },
    {
      "epoch": 17.771074710347584,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 44481
    },
    {
      "epoch": 17.77147423092289,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 44482
    },
    {
      "epoch": 17.7718737514982,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 44483
    },
    {
      "epoch": 17.772273272073512,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 44484
    },
    {
      "epoch": 17.772672792648823,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 44485
    },
    {
      "epoch": 17.77307231322413,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 44486
    },
    {
      "epoch": 17.77347183379944,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 44487
    },
    {
      "epoch": 17.77387135437475,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 44488
    },
    {
      "epoch": 17.77427087495006,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 44489
    },
    {
      "epoch": 17.77467039552537,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0079,
      "step": 44490
    },
    {
      "epoch": 17.77506991610068,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 44491
    },
    {
      "epoch": 17.775469436675987,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 44492
    },
    {
      "epoch": 17.775868957251298,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 44493
    },
    {
      "epoch": 17.77626847782661,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 44494
    },
    {
      "epoch": 17.77666799840192,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 44495
    },
    {
      "epoch": 17.777067518977226,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 44496
    },
    {
      "epoch": 17.777467039552537,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0089,
      "step": 44497
    },
    {
      "epoch": 17.777866560127848,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 44498
    },
    {
      "epoch": 17.778266080703155,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 44499
    },
    {
      "epoch": 17.778665601278465,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 44500
    },
    {
      "epoch": 17.779065121853776,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 44501
    },
    {
      "epoch": 17.779464642429087,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 44502
    },
    {
      "epoch": 17.779864163004394,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 44503
    },
    {
      "epoch": 17.780263683579705,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 44504
    },
    {
      "epoch": 17.780663204155015,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 44505
    },
    {
      "epoch": 17.781062724730322,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 44506
    },
    {
      "epoch": 17.781462245305633,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 44507
    },
    {
      "epoch": 17.781861765880944,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 44508
    },
    {
      "epoch": 17.78226128645625,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 44509
    },
    {
      "epoch": 17.78266080703156,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 44510
    },
    {
      "epoch": 17.783060327606872,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 44511
    },
    {
      "epoch": 17.783459848182183,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 44512
    },
    {
      "epoch": 17.78385936875749,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 44513
    },
    {
      "epoch": 17.7842588893328,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 44514
    },
    {
      "epoch": 17.78465840990811,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0124,
      "step": 44515
    },
    {
      "epoch": 17.78505793048342,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 44516
    },
    {
      "epoch": 17.78545745105873,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 44517
    },
    {
      "epoch": 17.78585697163404,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 44518
    },
    {
      "epoch": 17.78625649220935,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 44519
    },
    {
      "epoch": 17.786656012784658,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 44520
    },
    {
      "epoch": 17.78705553335997,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 44521
    },
    {
      "epoch": 17.78745505393528,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 44522
    },
    {
      "epoch": 17.787854574510586,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 44523
    },
    {
      "epoch": 17.788254095085897,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 44524
    },
    {
      "epoch": 17.788653615661207,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 44525
    },
    {
      "epoch": 17.789053136236515,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 44526
    },
    {
      "epoch": 17.789452656811825,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0404,
      "step": 44527
    },
    {
      "epoch": 17.789852177387136,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 44528
    },
    {
      "epoch": 17.790251697962447,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 44529
    },
    {
      "epoch": 17.790651218537754,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0116,
      "step": 44530
    },
    {
      "epoch": 17.791050739113064,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 44531
    },
    {
      "epoch": 17.791450259688375,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 44532
    },
    {
      "epoch": 17.791849780263682,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 44533
    },
    {
      "epoch": 17.792249300838993,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 44534
    },
    {
      "epoch": 17.792648821414303,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 44535
    },
    {
      "epoch": 17.793048341989614,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 44536
    },
    {
      "epoch": 17.79344786256492,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 44537
    },
    {
      "epoch": 17.793847383140232,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 44538
    },
    {
      "epoch": 17.794246903715543,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 44539
    },
    {
      "epoch": 17.79464642429085,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 44540
    },
    {
      "epoch": 17.79504594486616,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0113,
      "step": 44541
    },
    {
      "epoch": 17.79544546544147,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 44542
    },
    {
      "epoch": 17.795844986016778,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 44543
    },
    {
      "epoch": 17.79624450659209,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 44544
    },
    {
      "epoch": 17.7966440271674,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 44545
    },
    {
      "epoch": 17.79704354774271,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 44546
    },
    {
      "epoch": 17.797443068318017,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 44547
    },
    {
      "epoch": 17.797842588893328,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 44548
    },
    {
      "epoch": 17.79824210946864,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 44549
    },
    {
      "epoch": 17.798641630043946,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 44550
    },
    {
      "epoch": 17.799041150619257,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 44551
    },
    {
      "epoch": 17.799440671194567,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 44552
    },
    {
      "epoch": 17.799840191769878,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 44553
    },
    {
      "epoch": 17.800239712345185,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 44554
    },
    {
      "epoch": 17.800639232920496,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 44555
    },
    {
      "epoch": 17.801038753495806,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 44556
    },
    {
      "epoch": 17.801438274071113,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.043,
      "step": 44557
    },
    {
      "epoch": 17.801837794646424,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 44558
    },
    {
      "epoch": 17.802237315221735,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 44559
    },
    {
      "epoch": 17.802636835797042,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 44560
    },
    {
      "epoch": 17.803036356372353,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 44561
    },
    {
      "epoch": 17.803435876947663,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 44562
    },
    {
      "epoch": 17.803835397522974,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 44563
    },
    {
      "epoch": 17.80423491809828,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 44564
    },
    {
      "epoch": 17.80463443867359,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 44565
    },
    {
      "epoch": 17.805033959248902,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 44566
    },
    {
      "epoch": 17.80543347982421,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 44567
    },
    {
      "epoch": 17.80583300039952,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 44568
    },
    {
      "epoch": 17.80623252097483,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 44569
    },
    {
      "epoch": 17.806632041550138,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 44570
    },
    {
      "epoch": 17.80703156212545,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 44571
    },
    {
      "epoch": 17.80743108270076,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 44572
    },
    {
      "epoch": 17.80783060327607,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 44573
    },
    {
      "epoch": 17.808230123851377,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 44574
    },
    {
      "epoch": 17.808629644426688,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 44575
    },
    {
      "epoch": 17.809029165002,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 44576
    },
    {
      "epoch": 17.809428685577306,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 44577
    },
    {
      "epoch": 17.809828206152616,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 44578
    },
    {
      "epoch": 17.810227726727927,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 44579
    },
    {
      "epoch": 17.810627247303238,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 44580
    },
    {
      "epoch": 17.811026767878545,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 44581
    },
    {
      "epoch": 17.811426288453855,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 44582
    },
    {
      "epoch": 17.811825809029166,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 44583
    },
    {
      "epoch": 17.812225329604473,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0089,
      "step": 44584
    },
    {
      "epoch": 17.812624850179784,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 44585
    },
    {
      "epoch": 17.813024370755095,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 44586
    },
    {
      "epoch": 17.8134238913304,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 44587
    },
    {
      "epoch": 17.813823411905712,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 44588
    },
    {
      "epoch": 17.814222932481023,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 44589
    },
    {
      "epoch": 17.814622453056334,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 44590
    },
    {
      "epoch": 17.81502197363164,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 44591
    },
    {
      "epoch": 17.81542149420695,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 44592
    },
    {
      "epoch": 17.815821014782262,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 44593
    },
    {
      "epoch": 17.81622053535757,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 44594
    },
    {
      "epoch": 17.81662005593288,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 44595
    },
    {
      "epoch": 17.81701957650819,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 44596
    },
    {
      "epoch": 17.8174190970835,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 44597
    },
    {
      "epoch": 17.81781861765881,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 44598
    },
    {
      "epoch": 17.81821813823412,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 44599
    },
    {
      "epoch": 17.81861765880943,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 44600
    },
    {
      "epoch": 17.819017179384737,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 44601
    },
    {
      "epoch": 17.819416699960048,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 44602
    },
    {
      "epoch": 17.81981622053536,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 44603
    },
    {
      "epoch": 17.820215741110665,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 44604
    },
    {
      "epoch": 17.820615261685976,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 44605
    },
    {
      "epoch": 17.821014782261287,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 44606
    },
    {
      "epoch": 17.821414302836597,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 44607
    },
    {
      "epoch": 17.821813823411905,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0454,
      "step": 44608
    },
    {
      "epoch": 17.822213343987215,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 44609
    },
    {
      "epoch": 17.822612864562526,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 44610
    },
    {
      "epoch": 17.823012385137833,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 44611
    },
    {
      "epoch": 17.823411905713144,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 44612
    },
    {
      "epoch": 17.823811426288454,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 44613
    },
    {
      "epoch": 17.824210946863765,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 44614
    },
    {
      "epoch": 17.824610467439072,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 44615
    },
    {
      "epoch": 17.825009988014383,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 44616
    },
    {
      "epoch": 17.825409508589694,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 44617
    },
    {
      "epoch": 17.825809029165,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 44618
    },
    {
      "epoch": 17.82620854974031,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 44619
    },
    {
      "epoch": 17.826608070315622,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 44620
    },
    {
      "epoch": 17.82700759089093,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 44621
    },
    {
      "epoch": 17.82740711146624,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 44622
    },
    {
      "epoch": 17.82780663204155,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 44623
    },
    {
      "epoch": 17.82820615261686,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 44624
    },
    {
      "epoch": 17.82860567319217,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 44625
    },
    {
      "epoch": 17.82900519376748,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 44626
    },
    {
      "epoch": 17.82940471434279,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 44627
    },
    {
      "epoch": 17.829804234918097,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 44628
    },
    {
      "epoch": 17.830203755493407,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 44629
    },
    {
      "epoch": 17.830603276068718,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 44630
    },
    {
      "epoch": 17.83100279664403,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 44631
    },
    {
      "epoch": 17.831402317219336,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 44632
    },
    {
      "epoch": 17.831801837794647,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 44633
    },
    {
      "epoch": 17.832201358369957,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 44634
    },
    {
      "epoch": 17.832600878945264,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 44635
    },
    {
      "epoch": 17.833000399520575,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 44636
    },
    {
      "epoch": 17.833399920095886,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 44637
    },
    {
      "epoch": 17.833799440671193,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 44638
    },
    {
      "epoch": 17.834198961246504,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 44639
    },
    {
      "epoch": 17.834598481821814,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0416,
      "step": 44640
    },
    {
      "epoch": 17.834998002397125,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 44641
    },
    {
      "epoch": 17.835397522972432,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 44642
    },
    {
      "epoch": 17.835797043547743,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 44643
    },
    {
      "epoch": 17.836196564123053,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 44644
    },
    {
      "epoch": 17.83659608469836,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 44645
    },
    {
      "epoch": 17.83699560527367,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 44646
    },
    {
      "epoch": 17.837395125848982,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 44647
    },
    {
      "epoch": 17.837794646424292,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 44648
    },
    {
      "epoch": 17.8381941669996,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 44649
    },
    {
      "epoch": 17.83859368757491,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 44650
    },
    {
      "epoch": 17.83899320815022,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 44651
    },
    {
      "epoch": 17.839392728725528,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0362,
      "step": 44652
    },
    {
      "epoch": 17.83979224930084,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 44653
    },
    {
      "epoch": 17.84019176987615,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 44654
    },
    {
      "epoch": 17.840591290451457,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 44655
    },
    {
      "epoch": 17.840990811026767,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 44656
    },
    {
      "epoch": 17.841390331602078,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 44657
    },
    {
      "epoch": 17.84178985217739,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 44658
    },
    {
      "epoch": 17.842189372752696,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 44659
    },
    {
      "epoch": 17.842588893328006,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 44660
    },
    {
      "epoch": 17.842988413903317,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 44661
    },
    {
      "epoch": 17.843387934478624,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 44662
    },
    {
      "epoch": 17.843787455053935,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 44663
    },
    {
      "epoch": 17.844186975629245,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 44664
    },
    {
      "epoch": 17.844586496204556,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 44665
    },
    {
      "epoch": 17.844986016779863,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 44666
    },
    {
      "epoch": 17.845385537355174,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 44667
    },
    {
      "epoch": 17.845785057930485,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 44668
    },
    {
      "epoch": 17.84618457850579,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 44669
    },
    {
      "epoch": 17.846584099081102,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 44670
    },
    {
      "epoch": 17.846983619656413,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 44671
    },
    {
      "epoch": 17.84738314023172,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 44672
    },
    {
      "epoch": 17.84778266080703,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 44673
    },
    {
      "epoch": 17.84818218138234,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 44674
    },
    {
      "epoch": 17.848581701957652,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 44675
    },
    {
      "epoch": 17.84898122253296,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 44676
    },
    {
      "epoch": 17.84938074310827,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 44677
    },
    {
      "epoch": 17.84978026368358,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 44678
    },
    {
      "epoch": 17.850179784258888,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 44679
    },
    {
      "epoch": 17.8505793048342,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 44680
    },
    {
      "epoch": 17.85097882540951,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 44681
    },
    {
      "epoch": 17.85137834598482,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 44682
    },
    {
      "epoch": 17.851777866560127,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 44683
    },
    {
      "epoch": 17.852177387135438,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 44684
    },
    {
      "epoch": 17.85257690771075,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 44685
    },
    {
      "epoch": 17.852976428286055,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 44686
    },
    {
      "epoch": 17.853375948861366,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 44687
    },
    {
      "epoch": 17.853775469436677,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 44688
    },
    {
      "epoch": 17.854174990011984,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 44689
    },
    {
      "epoch": 17.854574510587295,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 44690
    },
    {
      "epoch": 17.854974031162605,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 44691
    },
    {
      "epoch": 17.855373551737916,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 44692
    },
    {
      "epoch": 17.855773072313223,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 44693
    },
    {
      "epoch": 17.856172592888534,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 44694
    },
    {
      "epoch": 17.856572113463844,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 44695
    },
    {
      "epoch": 17.85697163403915,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 44696
    },
    {
      "epoch": 17.857371154614462,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 44697
    },
    {
      "epoch": 17.857770675189773,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 44698
    },
    {
      "epoch": 17.858170195765084,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 44699
    },
    {
      "epoch": 17.85856971634039,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 44700
    },
    {
      "epoch": 17.8589692369157,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 44701
    },
    {
      "epoch": 17.859368757491012,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 44702
    },
    {
      "epoch": 17.85976827806632,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 44703
    },
    {
      "epoch": 17.86016779864163,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 44704
    },
    {
      "epoch": 17.86056731921694,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 44705
    },
    {
      "epoch": 17.860966839792248,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 44706
    },
    {
      "epoch": 17.86136636036756,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 44707
    },
    {
      "epoch": 17.86176588094287,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 44708
    },
    {
      "epoch": 17.86216540151818,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 44709
    },
    {
      "epoch": 17.862564922093487,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 44710
    },
    {
      "epoch": 17.862964442668797,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 44711
    },
    {
      "epoch": 17.863363963244108,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 44712
    },
    {
      "epoch": 17.863763483819415,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 44713
    },
    {
      "epoch": 17.864163004394726,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 44714
    },
    {
      "epoch": 17.864562524970037,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 44715
    },
    {
      "epoch": 17.864962045545347,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 44716
    },
    {
      "epoch": 17.865361566120654,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0064,
      "step": 44717
    },
    {
      "epoch": 17.865761086695965,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 44718
    },
    {
      "epoch": 17.866160607271276,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 44719
    },
    {
      "epoch": 17.866560127846583,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 44720
    },
    {
      "epoch": 17.866959648421894,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 44721
    },
    {
      "epoch": 17.867359168997204,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 44722
    },
    {
      "epoch": 17.86775868957251,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 44723
    },
    {
      "epoch": 17.868158210147822,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 44724
    },
    {
      "epoch": 17.868557730723133,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 44725
    },
    {
      "epoch": 17.868957251298443,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 44726
    },
    {
      "epoch": 17.86935677187375,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 44727
    },
    {
      "epoch": 17.86975629244906,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 44728
    },
    {
      "epoch": 17.870155813024372,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 44729
    },
    {
      "epoch": 17.87055533359968,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 44730
    },
    {
      "epoch": 17.87095485417499,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 44731
    },
    {
      "epoch": 17.8713543747503,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 44732
    },
    {
      "epoch": 17.87175389532561,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 44733
    },
    {
      "epoch": 17.872153415900918,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 44734
    },
    {
      "epoch": 17.87255293647623,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 44735
    },
    {
      "epoch": 17.87295245705154,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 44736
    },
    {
      "epoch": 17.873351977626847,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 44737
    },
    {
      "epoch": 17.873751498202157,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 44738
    },
    {
      "epoch": 17.874151018777468,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 44739
    },
    {
      "epoch": 17.874550539352775,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 44740
    },
    {
      "epoch": 17.874950059928086,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 44741
    },
    {
      "epoch": 17.875349580503396,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 44742
    },
    {
      "epoch": 17.875749101078707,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 44743
    },
    {
      "epoch": 17.876148621654014,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 44744
    },
    {
      "epoch": 17.876548142229325,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 44745
    },
    {
      "epoch": 17.876947662804636,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 44746
    },
    {
      "epoch": 17.877347183379943,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 44747
    },
    {
      "epoch": 17.877746703955253,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 44748
    },
    {
      "epoch": 17.878146224530564,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 44749
    },
    {
      "epoch": 17.878545745105875,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 44750
    },
    {
      "epoch": 17.878945265681182,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 44751
    },
    {
      "epoch": 17.879344786256492,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 44752
    },
    {
      "epoch": 17.879744306831803,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 44753
    },
    {
      "epoch": 17.88014382740711,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 44754
    },
    {
      "epoch": 17.88054334798242,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0117,
      "step": 44755
    },
    {
      "epoch": 17.88094286855773,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 44756
    },
    {
      "epoch": 17.88134238913304,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 44757
    },
    {
      "epoch": 17.88174190970835,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 44758
    },
    {
      "epoch": 17.88214143028366,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 44759
    },
    {
      "epoch": 17.88254095085897,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 44760
    },
    {
      "epoch": 17.882940471434278,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 44761
    },
    {
      "epoch": 17.88333999200959,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 44762
    },
    {
      "epoch": 17.8837395125849,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 44763
    },
    {
      "epoch": 17.884139033160206,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 44764
    },
    {
      "epoch": 17.884538553735517,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 44765
    },
    {
      "epoch": 17.884938074310828,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 44766
    },
    {
      "epoch": 17.88533759488614,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0069,
      "step": 44767
    },
    {
      "epoch": 17.885737115461446,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 44768
    },
    {
      "epoch": 17.886136636036756,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 44769
    },
    {
      "epoch": 17.886536156612067,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 44770
    },
    {
      "epoch": 17.886935677187374,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 44771
    },
    {
      "epoch": 17.887335197762685,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 44772
    },
    {
      "epoch": 17.887734718337995,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 44773
    },
    {
      "epoch": 17.888134238913302,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 44774
    },
    {
      "epoch": 17.888533759488613,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 44775
    },
    {
      "epoch": 17.888933280063924,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 44776
    },
    {
      "epoch": 17.889332800639234,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 44777
    },
    {
      "epoch": 17.88973232121454,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 44778
    },
    {
      "epoch": 17.890131841789852,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 44779
    },
    {
      "epoch": 17.890531362365163,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 44780
    },
    {
      "epoch": 17.89093088294047,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 44781
    },
    {
      "epoch": 17.89133040351578,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 44782
    },
    {
      "epoch": 17.89172992409109,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 44783
    },
    {
      "epoch": 17.892129444666402,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 44784
    },
    {
      "epoch": 17.89252896524171,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 44785
    },
    {
      "epoch": 17.89292848581702,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 44786
    },
    {
      "epoch": 17.89332800639233,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 44787
    },
    {
      "epoch": 17.893727526967638,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 44788
    },
    {
      "epoch": 17.89412704754295,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 44789
    },
    {
      "epoch": 17.89452656811826,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 44790
    },
    {
      "epoch": 17.894926088693566,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 44791
    },
    {
      "epoch": 17.895325609268877,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 44792
    },
    {
      "epoch": 17.895725129844188,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 44793
    },
    {
      "epoch": 17.896124650419498,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 44794
    },
    {
      "epoch": 17.896524170994805,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 44795
    },
    {
      "epoch": 17.896923691570116,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 44796
    },
    {
      "epoch": 17.897323212145427,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 44797
    },
    {
      "epoch": 17.897722732720734,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 44798
    },
    {
      "epoch": 17.898122253296044,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 44799
    },
    {
      "epoch": 17.898521773871355,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 44800
    },
    {
      "epoch": 17.898921294446666,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 44801
    },
    {
      "epoch": 17.899320815021973,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 44802
    },
    {
      "epoch": 17.899720335597284,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 44803
    },
    {
      "epoch": 17.900119856172594,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 44804
    },
    {
      "epoch": 17.9005193767479,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.008,
      "step": 44805
    },
    {
      "epoch": 17.900918897323212,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 44806
    },
    {
      "epoch": 17.901318417898523,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 44807
    },
    {
      "epoch": 17.90171793847383,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 44808
    },
    {
      "epoch": 17.90211745904914,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 44809
    },
    {
      "epoch": 17.90251697962445,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 44810
    },
    {
      "epoch": 17.902916500199762,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 44811
    },
    {
      "epoch": 17.90331602077507,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 44812
    },
    {
      "epoch": 17.90371554135038,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 44813
    },
    {
      "epoch": 17.90411506192569,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 44814
    },
    {
      "epoch": 17.904514582500997,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 44815
    },
    {
      "epoch": 17.904914103076308,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 44816
    },
    {
      "epoch": 17.90531362365162,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 44817
    },
    {
      "epoch": 17.90571314422693,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 44818
    },
    {
      "epoch": 17.906112664802237,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 44819
    },
    {
      "epoch": 17.906512185377547,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 44820
    },
    {
      "epoch": 17.906911705952858,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 44821
    },
    {
      "epoch": 17.907311226528165,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 44822
    },
    {
      "epoch": 17.907710747103476,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 44823
    },
    {
      "epoch": 17.908110267678786,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 44824
    },
    {
      "epoch": 17.908509788254094,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 44825
    },
    {
      "epoch": 17.908909308829404,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 44826
    },
    {
      "epoch": 17.909308829404715,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 44827
    },
    {
      "epoch": 17.909708349980026,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 44828
    },
    {
      "epoch": 17.910107870555333,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 44829
    },
    {
      "epoch": 17.910507391130643,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 44830
    },
    {
      "epoch": 17.910906911705954,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 44831
    },
    {
      "epoch": 17.91130643228126,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 44832
    },
    {
      "epoch": 17.911705952856572,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 44833
    },
    {
      "epoch": 17.912105473431883,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 44834
    },
    {
      "epoch": 17.91250499400719,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 44835
    },
    {
      "epoch": 17.9129045145825,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 44836
    },
    {
      "epoch": 17.91330403515781,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 44837
    },
    {
      "epoch": 17.91370355573312,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 44838
    },
    {
      "epoch": 17.91410307630843,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 44839
    },
    {
      "epoch": 17.91450259688374,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 44840
    },
    {
      "epoch": 17.91490211745905,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 44841
    },
    {
      "epoch": 17.915301638034357,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 44842
    },
    {
      "epoch": 17.915701158609668,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 44843
    },
    {
      "epoch": 17.91610067918498,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 44844
    },
    {
      "epoch": 17.91650019976029,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 44845
    },
    {
      "epoch": 17.916899720335596,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 44846
    },
    {
      "epoch": 17.917299240910907,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 44847
    },
    {
      "epoch": 17.917698761486218,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 44848
    },
    {
      "epoch": 17.918098282061525,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 44849
    },
    {
      "epoch": 17.918497802636836,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0118,
      "step": 44850
    },
    {
      "epoch": 17.918897323212146,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 44851
    },
    {
      "epoch": 17.919296843787453,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 44852
    },
    {
      "epoch": 17.919696364362764,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 44853
    },
    {
      "epoch": 17.920095884938075,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 44854
    },
    {
      "epoch": 17.920495405513385,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 44855
    },
    {
      "epoch": 17.920894926088693,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0113,
      "step": 44856
    },
    {
      "epoch": 17.921294446664003,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 44857
    },
    {
      "epoch": 17.921693967239314,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 44858
    },
    {
      "epoch": 17.92209348781462,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0112,
      "step": 44859
    },
    {
      "epoch": 17.92249300838993,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 44860
    },
    {
      "epoch": 17.922892528965242,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 44861
    },
    {
      "epoch": 17.923292049540553,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 44862
    },
    {
      "epoch": 17.92369157011586,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 44863
    },
    {
      "epoch": 17.92409109069117,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0082,
      "step": 44864
    },
    {
      "epoch": 17.92449061126648,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 44865
    },
    {
      "epoch": 17.92489013184179,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 44866
    },
    {
      "epoch": 17.9252896524171,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 44867
    },
    {
      "epoch": 17.92568917299241,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 44868
    },
    {
      "epoch": 17.926088693567717,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 44869
    },
    {
      "epoch": 17.926488214143028,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 44870
    },
    {
      "epoch": 17.92688773471834,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 44871
    },
    {
      "epoch": 17.92728725529365,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 44872
    },
    {
      "epoch": 17.927686775868956,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 44873
    },
    {
      "epoch": 17.928086296444267,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 44874
    },
    {
      "epoch": 17.928485817019578,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 44875
    },
    {
      "epoch": 17.928885337594885,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 44876
    },
    {
      "epoch": 17.929284858170195,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 44877
    },
    {
      "epoch": 17.929684378745506,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 44878
    },
    {
      "epoch": 17.930083899320817,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 44879
    },
    {
      "epoch": 17.930483419896124,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 44880
    },
    {
      "epoch": 17.930882940471434,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 44881
    },
    {
      "epoch": 17.931282461046745,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 44882
    },
    {
      "epoch": 17.931681981622052,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 44883
    },
    {
      "epoch": 17.932081502197363,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 44884
    },
    {
      "epoch": 17.932481022772674,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 44885
    },
    {
      "epoch": 17.93288054334798,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 44886
    },
    {
      "epoch": 17.93328006392329,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 44887
    },
    {
      "epoch": 17.933679584498602,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 44888
    },
    {
      "epoch": 17.934079105073913,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 44889
    },
    {
      "epoch": 17.93447862564922,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 44890
    },
    {
      "epoch": 17.93487814622453,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 44891
    },
    {
      "epoch": 17.93527766679984,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 44892
    },
    {
      "epoch": 17.93567718737515,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 44893
    },
    {
      "epoch": 17.93607670795046,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 44894
    },
    {
      "epoch": 17.93647622852577,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 44895
    },
    {
      "epoch": 17.93687574910108,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 44896
    },
    {
      "epoch": 17.937275269676388,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 44897
    },
    {
      "epoch": 17.937674790251698,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 44898
    },
    {
      "epoch": 17.93807431082701,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 44899
    },
    {
      "epoch": 17.938473831402316,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 44900
    },
    {
      "epoch": 17.938873351977627,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 44901
    },
    {
      "epoch": 17.939272872552937,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 44902
    },
    {
      "epoch": 17.939672393128244,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 44903
    },
    {
      "epoch": 17.940071913703555,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 44904
    },
    {
      "epoch": 17.940471434278866,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 44905
    },
    {
      "epoch": 17.940870954854176,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 44906
    },
    {
      "epoch": 17.941270475429484,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 44907
    },
    {
      "epoch": 17.941669996004794,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 44908
    },
    {
      "epoch": 17.942069516580105,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 44909
    },
    {
      "epoch": 17.942469037155412,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 44910
    },
    {
      "epoch": 17.942868557730723,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 44911
    },
    {
      "epoch": 17.943268078306033,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 44912
    },
    {
      "epoch": 17.943667598881344,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 44913
    },
    {
      "epoch": 17.94406711945665,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 44914
    },
    {
      "epoch": 17.944466640031962,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 44915
    },
    {
      "epoch": 17.944866160607273,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 44916
    },
    {
      "epoch": 17.94526568118258,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 44917
    },
    {
      "epoch": 17.94566520175789,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 44918
    },
    {
      "epoch": 17.9460647223332,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 44919
    },
    {
      "epoch": 17.946464242908508,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 44920
    },
    {
      "epoch": 17.94686376348382,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 44921
    },
    {
      "epoch": 17.94726328405913,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 44922
    },
    {
      "epoch": 17.94766280463444,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 44923
    },
    {
      "epoch": 17.948062325209747,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 44924
    },
    {
      "epoch": 17.948461845785058,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 44925
    },
    {
      "epoch": 17.94886136636037,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 44926
    },
    {
      "epoch": 17.949260886935676,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 44927
    },
    {
      "epoch": 17.949660407510986,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 44928
    },
    {
      "epoch": 17.950059928086297,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 44929
    },
    {
      "epoch": 17.950459448661604,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 44930
    },
    {
      "epoch": 17.950858969236915,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 44931
    },
    {
      "epoch": 17.951258489812226,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 44932
    },
    {
      "epoch": 17.951658010387536,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 44933
    },
    {
      "epoch": 17.952057530962843,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 44934
    },
    {
      "epoch": 17.952457051538154,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 44935
    },
    {
      "epoch": 17.952856572113465,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0092,
      "step": 44936
    },
    {
      "epoch": 17.953256092688772,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 44937
    },
    {
      "epoch": 17.953655613264083,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 44938
    },
    {
      "epoch": 17.954055133839393,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 44939
    },
    {
      "epoch": 17.954454654414704,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 44940
    },
    {
      "epoch": 17.95485417499001,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 44941
    },
    {
      "epoch": 17.95525369556532,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 44942
    },
    {
      "epoch": 17.955653216140632,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 44943
    },
    {
      "epoch": 17.95605273671594,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 44944
    },
    {
      "epoch": 17.95645225729125,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 44945
    },
    {
      "epoch": 17.95685177786656,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 44946
    },
    {
      "epoch": 17.957251298441868,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 44947
    },
    {
      "epoch": 17.95765081901718,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 44948
    },
    {
      "epoch": 17.95805033959249,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 44949
    },
    {
      "epoch": 17.9584498601678,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 44950
    },
    {
      "epoch": 17.958849380743107,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 44951
    },
    {
      "epoch": 17.959248901318418,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 44952
    },
    {
      "epoch": 17.95964842189373,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 44953
    },
    {
      "epoch": 17.960047942469036,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 44954
    },
    {
      "epoch": 17.960447463044346,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 44955
    },
    {
      "epoch": 17.960846983619657,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 44956
    },
    {
      "epoch": 17.961246504194968,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 44957
    },
    {
      "epoch": 17.961646024770275,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 44958
    },
    {
      "epoch": 17.962045545345585,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 44959
    },
    {
      "epoch": 17.962445065920896,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 44960
    },
    {
      "epoch": 17.962844586496203,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 44961
    },
    {
      "epoch": 17.963244107071514,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 44962
    },
    {
      "epoch": 17.963643627646825,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 44963
    },
    {
      "epoch": 17.96404314822213,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0091,
      "step": 44964
    },
    {
      "epoch": 17.964442668797442,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 44965
    },
    {
      "epoch": 17.964842189372753,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 44966
    },
    {
      "epoch": 17.965241709948064,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 44967
    },
    {
      "epoch": 17.96564123052337,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 44968
    },
    {
      "epoch": 17.96604075109868,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 44969
    },
    {
      "epoch": 17.966440271673992,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 44970
    },
    {
      "epoch": 17.9668397922493,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 44971
    },
    {
      "epoch": 17.96723931282461,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 44972
    },
    {
      "epoch": 17.96763883339992,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 44973
    },
    {
      "epoch": 17.96803835397523,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 44974
    },
    {
      "epoch": 17.96843787455054,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 44975
    },
    {
      "epoch": 17.96883739512585,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 44976
    },
    {
      "epoch": 17.96923691570116,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 44977
    },
    {
      "epoch": 17.969636436276467,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 44978
    },
    {
      "epoch": 17.970035956851778,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 44979
    },
    {
      "epoch": 17.97043547742709,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 44980
    },
    {
      "epoch": 17.970834998002395,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 44981
    },
    {
      "epoch": 17.971234518577706,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 44982
    },
    {
      "epoch": 17.971634039153017,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 44983
    },
    {
      "epoch": 17.972033559728327,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 44984
    },
    {
      "epoch": 17.972433080303635,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 44985
    },
    {
      "epoch": 17.972832600878945,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 44986
    },
    {
      "epoch": 17.973232121454256,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 44987
    },
    {
      "epoch": 17.973631642029563,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 44988
    },
    {
      "epoch": 17.974031162604874,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 44989
    },
    {
      "epoch": 17.974430683180184,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 44990
    },
    {
      "epoch": 17.974830203755495,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 44991
    },
    {
      "epoch": 17.975229724330802,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0417,
      "step": 44992
    },
    {
      "epoch": 17.975629244906113,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 44993
    },
    {
      "epoch": 17.976028765481423,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 44994
    },
    {
      "epoch": 17.97642828605673,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 44995
    },
    {
      "epoch": 17.97682780663204,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 44996
    },
    {
      "epoch": 17.977227327207352,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0396,
      "step": 44997
    },
    {
      "epoch": 17.97762684778266,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 44998
    },
    {
      "epoch": 17.97802636835797,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 44999
    },
    {
      "epoch": 17.97842588893328,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 45000
    },
    {
      "epoch": 17.97882540950859,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 45001
    },
    {
      "epoch": 17.979224930083898,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 45002
    },
    {
      "epoch": 17.97962445065921,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 45003
    },
    {
      "epoch": 17.98002397123452,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 45004
    },
    {
      "epoch": 17.980423491809827,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 45005
    },
    {
      "epoch": 17.980823012385137,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 45006
    },
    {
      "epoch": 17.981222532960448,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 45007
    },
    {
      "epoch": 17.98162205353576,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 45008
    },
    {
      "epoch": 17.982021574111066,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 45009
    },
    {
      "epoch": 17.982421094686377,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 45010
    },
    {
      "epoch": 17.982820615261687,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 45011
    },
    {
      "epoch": 17.983220135836994,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 45012
    },
    {
      "epoch": 17.983619656412305,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 45013
    },
    {
      "epoch": 17.984019176987616,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 45014
    },
    {
      "epoch": 17.984418697562923,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 45015
    },
    {
      "epoch": 17.984818218138233,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 45016
    },
    {
      "epoch": 17.985217738713544,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 45017
    },
    {
      "epoch": 17.985617259288855,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 45018
    },
    {
      "epoch": 17.986016779864162,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 45019
    },
    {
      "epoch": 17.986416300439473,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0104,
      "step": 45020
    },
    {
      "epoch": 17.986815821014783,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 45021
    },
    {
      "epoch": 17.98721534159009,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 45022
    },
    {
      "epoch": 17.9876148621654,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 45023
    },
    {
      "epoch": 17.98801438274071,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 45024
    },
    {
      "epoch": 17.988413903316022,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 45025
    },
    {
      "epoch": 17.98881342389133,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 45026
    },
    {
      "epoch": 17.98921294446664,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 45027
    },
    {
      "epoch": 17.98961246504195,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 45028
    },
    {
      "epoch": 17.990011985617258,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 45029
    },
    {
      "epoch": 17.99041150619257,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 45030
    },
    {
      "epoch": 17.99081102676788,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 45031
    },
    {
      "epoch": 17.991210547343186,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 45032
    },
    {
      "epoch": 17.991610067918497,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 45033
    },
    {
      "epoch": 17.992009588493808,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 45034
    },
    {
      "epoch": 17.99240910906912,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 45035
    },
    {
      "epoch": 17.992808629644426,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 45036
    },
    {
      "epoch": 17.993208150219736,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 45037
    },
    {
      "epoch": 17.993607670795047,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 45038
    },
    {
      "epoch": 17.994007191370354,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 45039
    },
    {
      "epoch": 17.994406711945665,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 45040
    },
    {
      "epoch": 17.994806232520975,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 45041
    },
    {
      "epoch": 17.995205753096286,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 45042
    },
    {
      "epoch": 17.995605273671593,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 45043
    },
    {
      "epoch": 17.996004794246904,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 45044
    },
    {
      "epoch": 17.996404314822215,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 45045
    },
    {
      "epoch": 17.99680383539752,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0388,
      "step": 45046
    },
    {
      "epoch": 17.997203355972832,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 45047
    },
    {
      "epoch": 17.997602876548143,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 45048
    },
    {
      "epoch": 17.99800239712345,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 45049
    },
    {
      "epoch": 17.99840191769876,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 45050
    },
    {
      "epoch": 17.99880143827407,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 45051
    },
    {
      "epoch": 17.999200958849382,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 45052
    },
    {
      "epoch": 17.99960047942469,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 45053
    },
    {
      "epoch": 18.0,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 45054
    },
    {
      "epoch": 18.00039952057531,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 45055
    },
    {
      "epoch": 18.000799041150618,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 45056
    },
    {
      "epoch": 18.00119856172593,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 45057
    },
    {
      "epoch": 18.00159808230124,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 45058
    },
    {
      "epoch": 18.00199760287655,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 45059
    },
    {
      "epoch": 18.002397123451857,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 45060
    },
    {
      "epoch": 18.002796644027168,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 45061
    },
    {
      "epoch": 18.00319616460248,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 45062
    },
    {
      "epoch": 18.003595685177785,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 45063
    },
    {
      "epoch": 18.003995205753096,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 45064
    },
    {
      "epoch": 18.004394726328407,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 45065
    },
    {
      "epoch": 18.004794246903714,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 45066
    },
    {
      "epoch": 18.005193767479025,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 45067
    },
    {
      "epoch": 18.005593288054335,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 45068
    },
    {
      "epoch": 18.005992808629646,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 45069
    },
    {
      "epoch": 18.006392329204953,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 45070
    },
    {
      "epoch": 18.006791849780264,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 45071
    },
    {
      "epoch": 18.007191370355574,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 45072
    },
    {
      "epoch": 18.00759089093088,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 45073
    },
    {
      "epoch": 18.007990411506192,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 45074
    },
    {
      "epoch": 18.008389932081503,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 45075
    },
    {
      "epoch": 18.008789452656814,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 45076
    },
    {
      "epoch": 18.00918897323212,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 45077
    },
    {
      "epoch": 18.00958849380743,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 45078
    },
    {
      "epoch": 18.009988014382742,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 45079
    },
    {
      "epoch": 18.01038753495805,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 45080
    },
    {
      "epoch": 18.01078705553336,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 45081
    },
    {
      "epoch": 18.01118657610867,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 45082
    },
    {
      "epoch": 18.011586096683978,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 45083
    },
    {
      "epoch": 18.01198561725929,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 45084
    },
    {
      "epoch": 18.0123851378346,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 45085
    },
    {
      "epoch": 18.01278465840991,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 45086
    },
    {
      "epoch": 18.013184178985217,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 45087
    },
    {
      "epoch": 18.013583699560527,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 45088
    },
    {
      "epoch": 18.013983220135838,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 45089
    },
    {
      "epoch": 18.014382740711145,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 45090
    },
    {
      "epoch": 18.014782261286456,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 45091
    },
    {
      "epoch": 18.015181781861767,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 45092
    },
    {
      "epoch": 18.015581302437077,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 45093
    },
    {
      "epoch": 18.015980823012384,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 45094
    },
    {
      "epoch": 18.016380343587695,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 45095
    },
    {
      "epoch": 18.016779864163006,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 45096
    },
    {
      "epoch": 18.017179384738313,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 45097
    },
    {
      "epoch": 18.017578905313623,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 45098
    },
    {
      "epoch": 18.017978425888934,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 45099
    },
    {
      "epoch": 18.01837794646424,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 45100
    },
    {
      "epoch": 18.018777467039552,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 45101
    },
    {
      "epoch": 18.019176987614863,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 45102
    },
    {
      "epoch": 18.019576508190173,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 45103
    },
    {
      "epoch": 18.01997602876548,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0105,
      "step": 45104
    },
    {
      "epoch": 18.02037554934079,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 45105
    },
    {
      "epoch": 18.020775069916102,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 45106
    },
    {
      "epoch": 18.02117459049141,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 45107
    },
    {
      "epoch": 18.02157411106672,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0108,
      "step": 45108
    },
    {
      "epoch": 18.02197363164203,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 45109
    },
    {
      "epoch": 18.02237315221734,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 45110
    },
    {
      "epoch": 18.022772672792648,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 45111
    },
    {
      "epoch": 18.02317219336796,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 45112
    },
    {
      "epoch": 18.02357171394327,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 45113
    },
    {
      "epoch": 18.023971234518577,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 45114
    },
    {
      "epoch": 18.024370755093887,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 45115
    },
    {
      "epoch": 18.024770275669198,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 45116
    },
    {
      "epoch": 18.025169796244505,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 45117
    },
    {
      "epoch": 18.025569316819816,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 45118
    },
    {
      "epoch": 18.025968837395126,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 45119
    },
    {
      "epoch": 18.026368357970437,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 45120
    },
    {
      "epoch": 18.026767878545744,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 45121
    },
    {
      "epoch": 18.027167399121055,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 45122
    },
    {
      "epoch": 18.027566919696365,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0401,
      "step": 45123
    },
    {
      "epoch": 18.027966440271673,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 45124
    },
    {
      "epoch": 18.028365960846983,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 45125
    },
    {
      "epoch": 18.028765481422294,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 45126
    },
    {
      "epoch": 18.029165001997605,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 45127
    },
    {
      "epoch": 18.02956452257291,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 45128
    },
    {
      "epoch": 18.029964043148222,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 45129
    },
    {
      "epoch": 18.030363563723533,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 45130
    },
    {
      "epoch": 18.03076308429884,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 45131
    },
    {
      "epoch": 18.03116260487415,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 45132
    },
    {
      "epoch": 18.03156212544946,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 45133
    },
    {
      "epoch": 18.03196164602477,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 45134
    },
    {
      "epoch": 18.03236116660008,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 45135
    },
    {
      "epoch": 18.03276068717539,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 45136
    },
    {
      "epoch": 18.0331602077507,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 45137
    },
    {
      "epoch": 18.033559728326008,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 45138
    },
    {
      "epoch": 18.03395924890132,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 45139
    },
    {
      "epoch": 18.03435876947663,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 45140
    },
    {
      "epoch": 18.034758290051936,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.01,
      "step": 45141
    },
    {
      "epoch": 18.035157810627247,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 45142
    },
    {
      "epoch": 18.035557331202558,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 45143
    },
    {
      "epoch": 18.03595685177787,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 45144
    },
    {
      "epoch": 18.036356372353175,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 45145
    },
    {
      "epoch": 18.036755892928486,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 45146
    },
    {
      "epoch": 18.037155413503797,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 45147
    },
    {
      "epoch": 18.037554934079104,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 45148
    },
    {
      "epoch": 18.037954454654415,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 45149
    },
    {
      "epoch": 18.038353975229725,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 45150
    },
    {
      "epoch": 18.038753495805032,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 45151
    },
    {
      "epoch": 18.039153016380343,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 45152
    },
    {
      "epoch": 18.039552536955654,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 45153
    },
    {
      "epoch": 18.039952057530964,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 45154
    },
    {
      "epoch": 18.04035157810627,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 45155
    },
    {
      "epoch": 18.040751098681582,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 45156
    },
    {
      "epoch": 18.041150619256893,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 45157
    },
    {
      "epoch": 18.0415501398322,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 45158
    },
    {
      "epoch": 18.04194966040751,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 45159
    },
    {
      "epoch": 18.04234918098282,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 45160
    },
    {
      "epoch": 18.04274870155813,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 45161
    },
    {
      "epoch": 18.04314822213344,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 45162
    },
    {
      "epoch": 18.04354774270875,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0452,
      "step": 45163
    },
    {
      "epoch": 18.04394726328406,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 45164
    },
    {
      "epoch": 18.044346783859368,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 45165
    },
    {
      "epoch": 18.04474630443468,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 45166
    },
    {
      "epoch": 18.04514582500999,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 45167
    },
    {
      "epoch": 18.045545345585296,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 45168
    },
    {
      "epoch": 18.045944866160607,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 45169
    },
    {
      "epoch": 18.046344386735917,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0106,
      "step": 45170
    },
    {
      "epoch": 18.046743907311228,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 45171
    },
    {
      "epoch": 18.047143427886535,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 45172
    },
    {
      "epoch": 18.047542948461846,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 45173
    },
    {
      "epoch": 18.047942469037157,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 45174
    },
    {
      "epoch": 18.048341989612464,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 45175
    },
    {
      "epoch": 18.048741510187774,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 45176
    },
    {
      "epoch": 18.049141030763085,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 45177
    },
    {
      "epoch": 18.049540551338392,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 45178
    },
    {
      "epoch": 18.049940071913703,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 45179
    },
    {
      "epoch": 18.050339592489014,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 45180
    },
    {
      "epoch": 18.050739113064324,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 45181
    },
    {
      "epoch": 18.05113863363963,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 45182
    },
    {
      "epoch": 18.051538154214942,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 45183
    },
    {
      "epoch": 18.051937674790253,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 45184
    },
    {
      "epoch": 18.05233719536556,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 45185
    },
    {
      "epoch": 18.05273671594087,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 45186
    },
    {
      "epoch": 18.05313623651618,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 45187
    },
    {
      "epoch": 18.053535757091492,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 45188
    },
    {
      "epoch": 18.0539352776668,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 45189
    },
    {
      "epoch": 18.05433479824211,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 45190
    },
    {
      "epoch": 18.05473431881742,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 45191
    },
    {
      "epoch": 18.055133839392727,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 45192
    },
    {
      "epoch": 18.055533359968038,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 45193
    },
    {
      "epoch": 18.05593288054335,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 45194
    },
    {
      "epoch": 18.056332401118656,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 45195
    },
    {
      "epoch": 18.056731921693967,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 45196
    },
    {
      "epoch": 18.057131442269277,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 45197
    },
    {
      "epoch": 18.057530962844588,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 45198
    },
    {
      "epoch": 18.057930483419895,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 45199
    },
    {
      "epoch": 18.058330003995206,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 45200
    },
    {
      "epoch": 18.058729524570516,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 45201
    },
    {
      "epoch": 18.059129045145824,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0124,
      "step": 45202
    },
    {
      "epoch": 18.059528565721134,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 45203
    },
    {
      "epoch": 18.059928086296445,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 45204
    },
    {
      "epoch": 18.060327606871756,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 45205
    },
    {
      "epoch": 18.060727127447063,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 45206
    },
    {
      "epoch": 18.061126648022373,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 45207
    },
    {
      "epoch": 18.061526168597684,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 45208
    },
    {
      "epoch": 18.06192568917299,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 45209
    },
    {
      "epoch": 18.062325209748302,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 45210
    },
    {
      "epoch": 18.062724730323612,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 45211
    },
    {
      "epoch": 18.06312425089892,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 45212
    },
    {
      "epoch": 18.06352377147423,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 45213
    },
    {
      "epoch": 18.06392329204954,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 45214
    },
    {
      "epoch": 18.06432281262485,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 45215
    },
    {
      "epoch": 18.06472233320016,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 45216
    },
    {
      "epoch": 18.06512185377547,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 45217
    },
    {
      "epoch": 18.06552137435078,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 45218
    },
    {
      "epoch": 18.065920894926087,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 45219
    },
    {
      "epoch": 18.066320415501398,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0109,
      "step": 45220
    },
    {
      "epoch": 18.06671993607671,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 45221
    },
    {
      "epoch": 18.06711945665202,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 45222
    },
    {
      "epoch": 18.067518977227326,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 45223
    },
    {
      "epoch": 18.067918497802637,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 45224
    },
    {
      "epoch": 18.068318018377948,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 45225
    },
    {
      "epoch": 18.068717538953255,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 45226
    },
    {
      "epoch": 18.069117059528566,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 45227
    },
    {
      "epoch": 18.069516580103876,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 45228
    },
    {
      "epoch": 18.069916100679183,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0109,
      "step": 45229
    },
    {
      "epoch": 18.070315621254494,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 45230
    },
    {
      "epoch": 18.070715141829805,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 45231
    },
    {
      "epoch": 18.071114662405115,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 45232
    },
    {
      "epoch": 18.071514182980422,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 45233
    },
    {
      "epoch": 18.071913703555733,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 45234
    },
    {
      "epoch": 18.072313224131044,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 45235
    },
    {
      "epoch": 18.07271274470635,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 45236
    },
    {
      "epoch": 18.07311226528166,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 45237
    },
    {
      "epoch": 18.073511785856972,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 45238
    },
    {
      "epoch": 18.073911306432283,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 45239
    },
    {
      "epoch": 18.07431082700759,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0118,
      "step": 45240
    },
    {
      "epoch": 18.0747103475829,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 45241
    },
    {
      "epoch": 18.07510986815821,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 45242
    },
    {
      "epoch": 18.07550938873352,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 45243
    },
    {
      "epoch": 18.07590890930883,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 45244
    },
    {
      "epoch": 18.07630842988414,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 45245
    },
    {
      "epoch": 18.076707950459447,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0128,
      "step": 45246
    },
    {
      "epoch": 18.077107471034758,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 45247
    },
    {
      "epoch": 18.07750699161007,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 45248
    },
    {
      "epoch": 18.07790651218538,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 45249
    },
    {
      "epoch": 18.078306032760686,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 45250
    },
    {
      "epoch": 18.078705553335997,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 45251
    },
    {
      "epoch": 18.079105073911307,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 45252
    },
    {
      "epoch": 18.079504594486615,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 45253
    },
    {
      "epoch": 18.079904115061925,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 45254
    },
    {
      "epoch": 18.080303635637236,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 45255
    },
    {
      "epoch": 18.080703156212547,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 45256
    },
    {
      "epoch": 18.081102676787854,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0112,
      "step": 45257
    },
    {
      "epoch": 18.081502197363164,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 45258
    },
    {
      "epoch": 18.081901717938475,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 45259
    },
    {
      "epoch": 18.082301238513782,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 45260
    },
    {
      "epoch": 18.082700759089093,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 45261
    },
    {
      "epoch": 18.083100279664404,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 45262
    },
    {
      "epoch": 18.08349980023971,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 45263
    },
    {
      "epoch": 18.08389932081502,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 45264
    },
    {
      "epoch": 18.084298841390332,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 45265
    },
    {
      "epoch": 18.084698361965643,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 45266
    },
    {
      "epoch": 18.08509788254095,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 45267
    },
    {
      "epoch": 18.08549740311626,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 45268
    },
    {
      "epoch": 18.08589692369157,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 45269
    },
    {
      "epoch": 18.08629644426688,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 45270
    },
    {
      "epoch": 18.08669596484219,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 45271
    },
    {
      "epoch": 18.0870954854175,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 45272
    },
    {
      "epoch": 18.08749500599281,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 45273
    },
    {
      "epoch": 18.087894526568117,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 45274
    },
    {
      "epoch": 18.088294047143428,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 45275
    },
    {
      "epoch": 18.08869356771874,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 45276
    },
    {
      "epoch": 18.089093088294046,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 45277
    },
    {
      "epoch": 18.089492608869357,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 45278
    },
    {
      "epoch": 18.089892129444667,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 45279
    },
    {
      "epoch": 18.090291650019974,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 45280
    },
    {
      "epoch": 18.090691170595285,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 45281
    },
    {
      "epoch": 18.091090691170596,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 45282
    },
    {
      "epoch": 18.091490211745906,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 45283
    },
    {
      "epoch": 18.091889732321214,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 45284
    },
    {
      "epoch": 18.092289252896524,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 45285
    },
    {
      "epoch": 18.092688773471835,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 45286
    },
    {
      "epoch": 18.093088294047142,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 45287
    },
    {
      "epoch": 18.093487814622453,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 45288
    },
    {
      "epoch": 18.093887335197763,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 45289
    },
    {
      "epoch": 18.094286855773074,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 45290
    },
    {
      "epoch": 18.09468637634838,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 45291
    },
    {
      "epoch": 18.095085896923692,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 45292
    },
    {
      "epoch": 18.095485417499003,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 45293
    },
    {
      "epoch": 18.09588493807431,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 45294
    },
    {
      "epoch": 18.09628445864962,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 45295
    },
    {
      "epoch": 18.09668397922493,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 45296
    },
    {
      "epoch": 18.097083499800238,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 45297
    },
    {
      "epoch": 18.09748302037555,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 45298
    },
    {
      "epoch": 18.09788254095086,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 45299
    },
    {
      "epoch": 18.09828206152617,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 45300
    },
    {
      "epoch": 18.098681582101477,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 45301
    },
    {
      "epoch": 18.099081102676788,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 45302
    },
    {
      "epoch": 18.0994806232521,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 45303
    },
    {
      "epoch": 18.099880143827406,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0047,
      "step": 45304
    },
    {
      "epoch": 18.100279664402716,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 45305
    },
    {
      "epoch": 18.100679184978027,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 45306
    },
    {
      "epoch": 18.101078705553338,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 45307
    },
    {
      "epoch": 18.101478226128645,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 45308
    },
    {
      "epoch": 18.101877746703956,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 45309
    },
    {
      "epoch": 18.102277267279266,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 45310
    },
    {
      "epoch": 18.102676787854573,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 45311
    },
    {
      "epoch": 18.103076308429884,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 45312
    },
    {
      "epoch": 18.103475829005195,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 45313
    },
    {
      "epoch": 18.103875349580502,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 45314
    },
    {
      "epoch": 18.104274870155812,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 45315
    },
    {
      "epoch": 18.104674390731123,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 45316
    },
    {
      "epoch": 18.105073911306434,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 45317
    },
    {
      "epoch": 18.10547343188174,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 45318
    },
    {
      "epoch": 18.10587295245705,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 45319
    },
    {
      "epoch": 18.106272473032362,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 45320
    },
    {
      "epoch": 18.10667199360767,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 45321
    },
    {
      "epoch": 18.10707151418298,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 45322
    },
    {
      "epoch": 18.10747103475829,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0075,
      "step": 45323
    },
    {
      "epoch": 18.107870555333598,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 45324
    },
    {
      "epoch": 18.10827007590891,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 45325
    },
    {
      "epoch": 18.10866959648422,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 45326
    },
    {
      "epoch": 18.10906911705953,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 45327
    },
    {
      "epoch": 18.109468637634837,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 45328
    },
    {
      "epoch": 18.109868158210148,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 45329
    },
    {
      "epoch": 18.11026767878546,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 45330
    },
    {
      "epoch": 18.110667199360766,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 45331
    },
    {
      "epoch": 18.111066719936076,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 45332
    },
    {
      "epoch": 18.111466240511387,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 45333
    },
    {
      "epoch": 18.111865761086698,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 45334
    },
    {
      "epoch": 18.112265281662005,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 45335
    },
    {
      "epoch": 18.112664802237315,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 45336
    },
    {
      "epoch": 18.113064322812626,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 45337
    },
    {
      "epoch": 18.113463843387933,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 45338
    },
    {
      "epoch": 18.113863363963244,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 45339
    },
    {
      "epoch": 18.114262884538554,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 45340
    },
    {
      "epoch": 18.11466240511386,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 45341
    },
    {
      "epoch": 18.115061925689172,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 45342
    },
    {
      "epoch": 18.115461446264483,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 45343
    },
    {
      "epoch": 18.115860966839794,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 45344
    },
    {
      "epoch": 18.1162604874151,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 45345
    },
    {
      "epoch": 18.11666000799041,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 45346
    },
    {
      "epoch": 18.117059528565722,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 45347
    },
    {
      "epoch": 18.11745904914103,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 45348
    },
    {
      "epoch": 18.11785856971634,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 45349
    },
    {
      "epoch": 18.11825809029165,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 45350
    },
    {
      "epoch": 18.11865761086696,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 45351
    },
    {
      "epoch": 18.11905713144227,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 45352
    },
    {
      "epoch": 18.11945665201758,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 45353
    },
    {
      "epoch": 18.11985617259289,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 45354
    },
    {
      "epoch": 18.120255693168197,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 45355
    },
    {
      "epoch": 18.120655213743508,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 45356
    },
    {
      "epoch": 18.121054734318818,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 45357
    },
    {
      "epoch": 18.121454254894125,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0424,
      "step": 45358
    },
    {
      "epoch": 18.121853775469436,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 45359
    },
    {
      "epoch": 18.122253296044747,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 45360
    },
    {
      "epoch": 18.122652816620057,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 45361
    },
    {
      "epoch": 18.123052337195364,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 45362
    },
    {
      "epoch": 18.123451857770675,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 45363
    },
    {
      "epoch": 18.123851378345986,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 45364
    },
    {
      "epoch": 18.124250898921293,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 45365
    },
    {
      "epoch": 18.124650419496604,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 45366
    },
    {
      "epoch": 18.125049940071914,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 45367
    },
    {
      "epoch": 18.125449460647225,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 45368
    },
    {
      "epoch": 18.125848981222532,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 45369
    },
    {
      "epoch": 18.126248501797843,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 45370
    },
    {
      "epoch": 18.126648022373153,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 45371
    },
    {
      "epoch": 18.12704754294846,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 45372
    },
    {
      "epoch": 18.12744706352377,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0128,
      "step": 45373
    },
    {
      "epoch": 18.127846584099082,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 45374
    },
    {
      "epoch": 18.12824610467439,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 45375
    },
    {
      "epoch": 18.1286456252497,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 45376
    },
    {
      "epoch": 18.12904514582501,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 45377
    },
    {
      "epoch": 18.12944466640032,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 45378
    },
    {
      "epoch": 18.129844186975628,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 45379
    },
    {
      "epoch": 18.13024370755094,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 45380
    },
    {
      "epoch": 18.13064322812625,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 45381
    },
    {
      "epoch": 18.131042748701557,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 45382
    },
    {
      "epoch": 18.131442269276867,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 45383
    },
    {
      "epoch": 18.131841789852178,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 45384
    },
    {
      "epoch": 18.13224131042749,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 45385
    },
    {
      "epoch": 18.132640831002796,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 45386
    },
    {
      "epoch": 18.133040351578106,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 45387
    },
    {
      "epoch": 18.133439872153417,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 45388
    },
    {
      "epoch": 18.133839392728724,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 45389
    },
    {
      "epoch": 18.134238913304035,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 45390
    },
    {
      "epoch": 18.134638433879346,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 45391
    },
    {
      "epoch": 18.135037954454653,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 45392
    },
    {
      "epoch": 18.135437475029963,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 45393
    },
    {
      "epoch": 18.135836995605274,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 45394
    },
    {
      "epoch": 18.136236516180585,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0411,
      "step": 45395
    },
    {
      "epoch": 18.136636036755892,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 45396
    },
    {
      "epoch": 18.137035557331203,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 45397
    },
    {
      "epoch": 18.137435077906513,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 45398
    },
    {
      "epoch": 18.13783459848182,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 45399
    },
    {
      "epoch": 18.13823411905713,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0406,
      "step": 45400
    },
    {
      "epoch": 18.13863363963244,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 45401
    },
    {
      "epoch": 18.139033160207752,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 45402
    },
    {
      "epoch": 18.13943268078306,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 45403
    },
    {
      "epoch": 18.13983220135837,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 45404
    },
    {
      "epoch": 18.14023172193368,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 45405
    },
    {
      "epoch": 18.140631242508988,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 45406
    },
    {
      "epoch": 18.1410307630843,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 45407
    },
    {
      "epoch": 18.14143028365961,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 45408
    },
    {
      "epoch": 18.141829804234916,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 45409
    },
    {
      "epoch": 18.142229324810227,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 45410
    },
    {
      "epoch": 18.142628845385538,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 45411
    },
    {
      "epoch": 18.14302836596085,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 45412
    },
    {
      "epoch": 18.143427886536156,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 45413
    },
    {
      "epoch": 18.143827407111466,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 45414
    },
    {
      "epoch": 18.144226927686777,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 45415
    },
    {
      "epoch": 18.144626448262084,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 45416
    },
    {
      "epoch": 18.145025968837395,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 45417
    },
    {
      "epoch": 18.145425489412705,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 45418
    },
    {
      "epoch": 18.145825009988016,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 45419
    },
    {
      "epoch": 18.146224530563323,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 45420
    },
    {
      "epoch": 18.146624051138634,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 45421
    },
    {
      "epoch": 18.147023571713945,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 45422
    },
    {
      "epoch": 18.14742309228925,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 45423
    },
    {
      "epoch": 18.147822612864562,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 45424
    },
    {
      "epoch": 18.148222133439873,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 45425
    },
    {
      "epoch": 18.14862165401518,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 45426
    },
    {
      "epoch": 18.14902117459049,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 45427
    },
    {
      "epoch": 18.1494206951658,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 45428
    },
    {
      "epoch": 18.149820215741112,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 45429
    },
    {
      "epoch": 18.15021973631642,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 45430
    },
    {
      "epoch": 18.15061925689173,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 45431
    },
    {
      "epoch": 18.15101877746704,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 45432
    },
    {
      "epoch": 18.151418298042348,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 45433
    },
    {
      "epoch": 18.15181781861766,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 45434
    },
    {
      "epoch": 18.15221733919297,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 45435
    },
    {
      "epoch": 18.15261685976828,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 45436
    },
    {
      "epoch": 18.153016380343587,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 45437
    },
    {
      "epoch": 18.153415900918898,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 45438
    },
    {
      "epoch": 18.15381542149421,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 45439
    },
    {
      "epoch": 18.154214942069515,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 45440
    },
    {
      "epoch": 18.154614462644826,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 45441
    },
    {
      "epoch": 18.155013983220137,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 45442
    },
    {
      "epoch": 18.155413503795444,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 45443
    },
    {
      "epoch": 18.155813024370755,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 45444
    },
    {
      "epoch": 18.156212544946065,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 45445
    },
    {
      "epoch": 18.156612065521376,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 45446
    },
    {
      "epoch": 18.157011586096683,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 45447
    },
    {
      "epoch": 18.157411106671994,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 45448
    },
    {
      "epoch": 18.157810627247304,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0093,
      "step": 45449
    },
    {
      "epoch": 18.15821014782261,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 45450
    },
    {
      "epoch": 18.158609668397922,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 45451
    },
    {
      "epoch": 18.159009188973233,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 45452
    },
    {
      "epoch": 18.159408709548543,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 45453
    },
    {
      "epoch": 18.15980823012385,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 45454
    },
    {
      "epoch": 18.16020775069916,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 45455
    },
    {
      "epoch": 18.160607271274472,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 45456
    },
    {
      "epoch": 18.16100679184978,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 45457
    },
    {
      "epoch": 18.16140631242509,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 45458
    },
    {
      "epoch": 18.1618058330004,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 45459
    },
    {
      "epoch": 18.162205353575708,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 45460
    },
    {
      "epoch": 18.162604874151018,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0076,
      "step": 45461
    },
    {
      "epoch": 18.16300439472633,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 45462
    },
    {
      "epoch": 18.16340391530164,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 45463
    },
    {
      "epoch": 18.163803435876947,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 45464
    },
    {
      "epoch": 18.164202956452257,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0103,
      "step": 45465
    },
    {
      "epoch": 18.164602477027568,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 45466
    },
    {
      "epoch": 18.165001997602875,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 45467
    },
    {
      "epoch": 18.165401518178186,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 45468
    },
    {
      "epoch": 18.165801038753496,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 45469
    },
    {
      "epoch": 18.166200559328807,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 45470
    },
    {
      "epoch": 18.166600079904114,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 45471
    },
    {
      "epoch": 18.166999600479425,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 45472
    },
    {
      "epoch": 18.167399121054736,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 45473
    },
    {
      "epoch": 18.167798641630043,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 45474
    },
    {
      "epoch": 18.168198162205353,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 45475
    },
    {
      "epoch": 18.168597682780664,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 45476
    },
    {
      "epoch": 18.16899720335597,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 45477
    },
    {
      "epoch": 18.169396723931282,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 45478
    },
    {
      "epoch": 18.169796244506593,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 45479
    },
    {
      "epoch": 18.170195765081903,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 45480
    },
    {
      "epoch": 18.17059528565721,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 45481
    },
    {
      "epoch": 18.17099480623252,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 45482
    },
    {
      "epoch": 18.17139432680783,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 45483
    },
    {
      "epoch": 18.17179384738314,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 45484
    },
    {
      "epoch": 18.17219336795845,
      "grad_norm": 0.035888671875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 45485
    },
    {
      "epoch": 18.17259288853376,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 45486
    },
    {
      "epoch": 18.17299240910907,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 45487
    },
    {
      "epoch": 18.173391929684378,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 45488
    },
    {
      "epoch": 18.17379145025969,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 45489
    },
    {
      "epoch": 18.174190970835,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 45490
    },
    {
      "epoch": 18.174590491410306,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 45491
    },
    {
      "epoch": 18.174990011985617,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 45492
    },
    {
      "epoch": 18.175389532560928,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 45493
    },
    {
      "epoch": 18.175789053136235,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 45494
    },
    {
      "epoch": 18.176188573711546,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 45495
    },
    {
      "epoch": 18.176588094286856,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 45496
    },
    {
      "epoch": 18.176987614862167,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 45497
    },
    {
      "epoch": 18.177387135437474,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 45498
    },
    {
      "epoch": 18.177786656012785,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 45499
    },
    {
      "epoch": 18.178186176588095,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 45500
    },
    {
      "epoch": 18.178585697163403,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 45501
    },
    {
      "epoch": 18.178985217738713,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 45502
    },
    {
      "epoch": 18.179384738314024,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 45503
    },
    {
      "epoch": 18.179784258889335,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 45504
    },
    {
      "epoch": 18.18018377946464,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 45505
    },
    {
      "epoch": 18.180583300039952,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 45506
    },
    {
      "epoch": 18.180982820615263,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 45507
    },
    {
      "epoch": 18.18138234119057,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 45508
    },
    {
      "epoch": 18.18178186176588,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 45509
    },
    {
      "epoch": 18.18218138234119,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 45510
    },
    {
      "epoch": 18.1825809029165,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 45511
    },
    {
      "epoch": 18.18298042349181,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 45512
    },
    {
      "epoch": 18.18337994406712,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 45513
    },
    {
      "epoch": 18.18377946464243,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 45514
    },
    {
      "epoch": 18.184178985217738,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 45515
    },
    {
      "epoch": 18.18457850579305,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 45516
    },
    {
      "epoch": 18.18497802636836,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 45517
    },
    {
      "epoch": 18.185377546943666,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 45518
    },
    {
      "epoch": 18.185777067518977,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 45519
    },
    {
      "epoch": 18.186176588094288,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 45520
    },
    {
      "epoch": 18.186576108669595,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0055,
      "step": 45521
    },
    {
      "epoch": 18.186975629244905,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 45522
    },
    {
      "epoch": 18.187375149820216,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 45523
    },
    {
      "epoch": 18.187774670395527,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 45524
    },
    {
      "epoch": 18.188174190970834,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 45525
    },
    {
      "epoch": 18.188573711546145,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 45526
    },
    {
      "epoch": 18.188973232121455,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 45527
    },
    {
      "epoch": 18.189372752696762,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 45528
    },
    {
      "epoch": 18.189772273272073,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 45529
    },
    {
      "epoch": 18.190171793847384,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 45530
    },
    {
      "epoch": 18.190571314422694,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 45531
    },
    {
      "epoch": 18.190970834998,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 45532
    },
    {
      "epoch": 18.191370355573312,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 45533
    },
    {
      "epoch": 18.191769876148623,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 45534
    },
    {
      "epoch": 18.19216939672393,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 45535
    },
    {
      "epoch": 18.19256891729924,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 45536
    },
    {
      "epoch": 18.19296843787455,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 45537
    },
    {
      "epoch": 18.19336795844986,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 45538
    },
    {
      "epoch": 18.19376747902517,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 45539
    },
    {
      "epoch": 18.19416699960048,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 45540
    },
    {
      "epoch": 18.19456652017579,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 45541
    },
    {
      "epoch": 18.194966040751098,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 45542
    },
    {
      "epoch": 18.19536556132641,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 45543
    },
    {
      "epoch": 18.19576508190172,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 45544
    },
    {
      "epoch": 18.196164602477026,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 45545
    },
    {
      "epoch": 18.196564123052337,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 45546
    },
    {
      "epoch": 18.196963643627647,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 45547
    },
    {
      "epoch": 18.197363164202958,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 45548
    },
    {
      "epoch": 18.197762684778265,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 45549
    },
    {
      "epoch": 18.198162205353576,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 45550
    },
    {
      "epoch": 18.198561725928887,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 45551
    },
    {
      "epoch": 18.198961246504194,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 45552
    },
    {
      "epoch": 18.199360767079504,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 45553
    },
    {
      "epoch": 18.199760287654815,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 45554
    },
    {
      "epoch": 18.200159808230122,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 45555
    },
    {
      "epoch": 18.200559328805433,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 45556
    },
    {
      "epoch": 18.200958849380743,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 45557
    },
    {
      "epoch": 18.201358369956054,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 45558
    },
    {
      "epoch": 18.20175789053136,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0124,
      "step": 45559
    },
    {
      "epoch": 18.202157411106672,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 45560
    },
    {
      "epoch": 18.202556931681983,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 45561
    },
    {
      "epoch": 18.20295645225729,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 45562
    },
    {
      "epoch": 18.2033559728326,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 45563
    },
    {
      "epoch": 18.20375549340791,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 45564
    },
    {
      "epoch": 18.204155013983222,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 45565
    },
    {
      "epoch": 18.20455453455853,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 45566
    },
    {
      "epoch": 18.20495405513384,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 45567
    },
    {
      "epoch": 18.20535357570915,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 45568
    },
    {
      "epoch": 18.205753096284457,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 45569
    },
    {
      "epoch": 18.206152616859768,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 45570
    },
    {
      "epoch": 18.20655213743508,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0126,
      "step": 45571
    },
    {
      "epoch": 18.206951658010386,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 45572
    },
    {
      "epoch": 18.207351178585697,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0122,
      "step": 45573
    },
    {
      "epoch": 18.207750699161007,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 45574
    },
    {
      "epoch": 18.208150219736318,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 45575
    },
    {
      "epoch": 18.208549740311625,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 45576
    },
    {
      "epoch": 18.208949260886936,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 45577
    },
    {
      "epoch": 18.209348781462246,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 45578
    },
    {
      "epoch": 18.209748302037553,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 45579
    },
    {
      "epoch": 18.210147822612864,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 45580
    },
    {
      "epoch": 18.210547343188175,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 45581
    },
    {
      "epoch": 18.210946863763485,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0376,
      "step": 45582
    },
    {
      "epoch": 18.211346384338793,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 45583
    },
    {
      "epoch": 18.211745904914103,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 45584
    },
    {
      "epoch": 18.212145425489414,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 45585
    },
    {
      "epoch": 18.21254494606472,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 45586
    },
    {
      "epoch": 18.21294446664003,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 45587
    },
    {
      "epoch": 18.213343987215342,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 45588
    },
    {
      "epoch": 18.21374350779065,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 45589
    },
    {
      "epoch": 18.21414302836596,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0125,
      "step": 45590
    },
    {
      "epoch": 18.21454254894127,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 45591
    },
    {
      "epoch": 18.21494206951658,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 45592
    },
    {
      "epoch": 18.21534159009189,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 45593
    },
    {
      "epoch": 18.2157411106672,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 45594
    },
    {
      "epoch": 18.21614063124251,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 45595
    },
    {
      "epoch": 18.216540151817817,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 45596
    },
    {
      "epoch": 18.216939672393128,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 45597
    },
    {
      "epoch": 18.21733919296844,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 45598
    },
    {
      "epoch": 18.21773871354375,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 45599
    },
    {
      "epoch": 18.218138234119056,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 45600
    },
    {
      "epoch": 18.218537754694367,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 45601
    },
    {
      "epoch": 18.218937275269678,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 45602
    },
    {
      "epoch": 18.219336795844985,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 45603
    },
    {
      "epoch": 18.219736316420295,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 45604
    },
    {
      "epoch": 18.220135836995606,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 45605
    },
    {
      "epoch": 18.220535357570913,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 45606
    },
    {
      "epoch": 18.220934878146224,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 45607
    },
    {
      "epoch": 18.221334398721535,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 45608
    },
    {
      "epoch": 18.221733919296845,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0393,
      "step": 45609
    },
    {
      "epoch": 18.222133439872152,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0407,
      "step": 45610
    },
    {
      "epoch": 18.222532960447463,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 45611
    },
    {
      "epoch": 18.222932481022774,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 45612
    },
    {
      "epoch": 18.22333200159808,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 45613
    },
    {
      "epoch": 18.22373152217339,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 45614
    },
    {
      "epoch": 18.224131042748702,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 45615
    },
    {
      "epoch": 18.224530563324013,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 45616
    },
    {
      "epoch": 18.22493008389932,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 45617
    },
    {
      "epoch": 18.22532960447463,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 45618
    },
    {
      "epoch": 18.22572912504994,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 45619
    },
    {
      "epoch": 18.22612864562525,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 45620
    },
    {
      "epoch": 18.22652816620056,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 45621
    },
    {
      "epoch": 18.22692768677587,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0099,
      "step": 45622
    },
    {
      "epoch": 18.227327207351177,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 45623
    },
    {
      "epoch": 18.227726727926488,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 45624
    },
    {
      "epoch": 18.2281262485018,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 45625
    },
    {
      "epoch": 18.22852576907711,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 45626
    },
    {
      "epoch": 18.228925289652416,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 45627
    },
    {
      "epoch": 18.229324810227727,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 45628
    },
    {
      "epoch": 18.229724330803037,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 45629
    },
    {
      "epoch": 18.230123851378345,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 45630
    },
    {
      "epoch": 18.230523371953655,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 45631
    },
    {
      "epoch": 18.230922892528966,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 45632
    },
    {
      "epoch": 18.231322413104277,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 45633
    },
    {
      "epoch": 18.231721933679584,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 45634
    },
    {
      "epoch": 18.232121454254894,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 45635
    },
    {
      "epoch": 18.232520974830205,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 45636
    },
    {
      "epoch": 18.232920495405512,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 45637
    },
    {
      "epoch": 18.233320015980823,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 45638
    },
    {
      "epoch": 18.233719536556134,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 45639
    },
    {
      "epoch": 18.23411905713144,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 45640
    },
    {
      "epoch": 18.23451857770675,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 45641
    },
    {
      "epoch": 18.234918098282062,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 45642
    },
    {
      "epoch": 18.235317618857373,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 45643
    },
    {
      "epoch": 18.23571713943268,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 45644
    },
    {
      "epoch": 18.23611666000799,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 45645
    },
    {
      "epoch": 18.2365161805833,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 45646
    },
    {
      "epoch": 18.23691570115861,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0134,
      "step": 45647
    },
    {
      "epoch": 18.23731522173392,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 45648
    },
    {
      "epoch": 18.23771474230923,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 45649
    },
    {
      "epoch": 18.23811426288454,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 45650
    },
    {
      "epoch": 18.238513783459847,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0377,
      "step": 45651
    },
    {
      "epoch": 18.238913304035158,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 45652
    },
    {
      "epoch": 18.23931282461047,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 45653
    },
    {
      "epoch": 18.239712345185776,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 45654
    },
    {
      "epoch": 18.240111865761087,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 45655
    },
    {
      "epoch": 18.240511386336397,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 45656
    },
    {
      "epoch": 18.240910906911704,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 45657
    },
    {
      "epoch": 18.241310427487015,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 45658
    },
    {
      "epoch": 18.241709948062326,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 45659
    },
    {
      "epoch": 18.242109468637636,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 45660
    },
    {
      "epoch": 18.242508989212944,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 45661
    },
    {
      "epoch": 18.242908509788254,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 45662
    },
    {
      "epoch": 18.243308030363565,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 45663
    },
    {
      "epoch": 18.243707550938872,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 45664
    },
    {
      "epoch": 18.244107071514183,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 45665
    },
    {
      "epoch": 18.244506592089493,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 45666
    },
    {
      "epoch": 18.244906112664804,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 45667
    },
    {
      "epoch": 18.24530563324011,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 45668
    },
    {
      "epoch": 18.245705153815422,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 45669
    },
    {
      "epoch": 18.246104674390732,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 45670
    },
    {
      "epoch": 18.24650419496604,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 45671
    },
    {
      "epoch": 18.24690371554135,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 45672
    },
    {
      "epoch": 18.24730323611666,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 45673
    },
    {
      "epoch": 18.247702756691968,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 45674
    },
    {
      "epoch": 18.24810227726728,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 45675
    },
    {
      "epoch": 18.24850179784259,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 45676
    },
    {
      "epoch": 18.2489013184179,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 45677
    },
    {
      "epoch": 18.249300838993207,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 45678
    },
    {
      "epoch": 18.249700359568518,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 45679
    },
    {
      "epoch": 18.25009988014383,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 45680
    },
    {
      "epoch": 18.250499400719136,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 45681
    },
    {
      "epoch": 18.250898921294446,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 45682
    },
    {
      "epoch": 18.251298441869757,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 45683
    },
    {
      "epoch": 18.251697962445064,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 45684
    },
    {
      "epoch": 18.252097483020375,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 45685
    },
    {
      "epoch": 18.252497003595685,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 45686
    },
    {
      "epoch": 18.252896524170996,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 45687
    },
    {
      "epoch": 18.253296044746303,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 45688
    },
    {
      "epoch": 18.253695565321614,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0118,
      "step": 45689
    },
    {
      "epoch": 18.254095085896925,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 45690
    },
    {
      "epoch": 18.25449460647223,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0124,
      "step": 45691
    },
    {
      "epoch": 18.254894127047542,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 45692
    },
    {
      "epoch": 18.255293647622853,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 45693
    },
    {
      "epoch": 18.255693168198164,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 45694
    },
    {
      "epoch": 18.25609268877347,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 45695
    },
    {
      "epoch": 18.25649220934878,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 45696
    },
    {
      "epoch": 18.256891729924092,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 45697
    },
    {
      "epoch": 18.2572912504994,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 45698
    },
    {
      "epoch": 18.25769077107471,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 45699
    },
    {
      "epoch": 18.25809029165002,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 45700
    },
    {
      "epoch": 18.258489812225328,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 45701
    },
    {
      "epoch": 18.25888933280064,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 45702
    },
    {
      "epoch": 18.25928885337595,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 45703
    },
    {
      "epoch": 18.25968837395126,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 45704
    },
    {
      "epoch": 18.260087894526567,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 45705
    },
    {
      "epoch": 18.260487415101878,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 45706
    },
    {
      "epoch": 18.26088693567719,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 45707
    },
    {
      "epoch": 18.261286456252495,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 45708
    },
    {
      "epoch": 18.261685976827806,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 45709
    },
    {
      "epoch": 18.262085497403117,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 45710
    },
    {
      "epoch": 18.262485017978427,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 45711
    },
    {
      "epoch": 18.262884538553735,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 45712
    },
    {
      "epoch": 18.263284059129045,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 45713
    },
    {
      "epoch": 18.263683579704356,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 45714
    },
    {
      "epoch": 18.264083100279663,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 45715
    },
    {
      "epoch": 18.264482620854974,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 45716
    },
    {
      "epoch": 18.264882141430284,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 45717
    },
    {
      "epoch": 18.26528166200559,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 45718
    },
    {
      "epoch": 18.265681182580902,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 45719
    },
    {
      "epoch": 18.266080703156213,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 45720
    },
    {
      "epoch": 18.266480223731524,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 45721
    },
    {
      "epoch": 18.26687974430683,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0399,
      "step": 45722
    },
    {
      "epoch": 18.26727926488214,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 45723
    },
    {
      "epoch": 18.267678785457452,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 45724
    },
    {
      "epoch": 18.26807830603276,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 45725
    },
    {
      "epoch": 18.26847782660807,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 45726
    },
    {
      "epoch": 18.26887734718338,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 45727
    },
    {
      "epoch": 18.26927686775869,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 45728
    },
    {
      "epoch": 18.269676388334,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 45729
    },
    {
      "epoch": 18.27007590890931,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 45730
    },
    {
      "epoch": 18.27047542948462,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 45731
    },
    {
      "epoch": 18.270874950059927,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 45732
    },
    {
      "epoch": 18.271274470635237,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 45733
    },
    {
      "epoch": 18.271673991210548,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 45734
    },
    {
      "epoch": 18.272073511785855,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 45735
    },
    {
      "epoch": 18.272473032361166,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 45736
    },
    {
      "epoch": 18.272872552936477,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 45737
    },
    {
      "epoch": 18.273272073511787,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 45738
    },
    {
      "epoch": 18.273671594087094,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 45739
    },
    {
      "epoch": 18.274071114662405,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 45740
    },
    {
      "epoch": 18.274470635237716,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 45741
    },
    {
      "epoch": 18.274870155813023,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 45742
    },
    {
      "epoch": 18.275269676388334,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 45743
    },
    {
      "epoch": 18.275669196963644,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 45744
    },
    {
      "epoch": 18.276068717538955,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 45745
    },
    {
      "epoch": 18.276468238114262,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 45746
    },
    {
      "epoch": 18.276867758689573,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0312,
      "step": 45747
    },
    {
      "epoch": 18.277267279264883,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 45748
    },
    {
      "epoch": 18.27766679984019,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 45749
    },
    {
      "epoch": 18.2780663204155,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 45750
    },
    {
      "epoch": 18.278465840990812,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 45751
    },
    {
      "epoch": 18.27886536156612,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 45752
    },
    {
      "epoch": 18.27926488214143,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0102,
      "step": 45753
    },
    {
      "epoch": 18.27966440271674,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 45754
    },
    {
      "epoch": 18.28006392329205,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 45755
    },
    {
      "epoch": 18.280463443867358,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 45756
    },
    {
      "epoch": 18.28086296444267,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 45757
    },
    {
      "epoch": 18.28126248501798,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 45758
    },
    {
      "epoch": 18.281662005593287,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 45759
    },
    {
      "epoch": 18.282061526168597,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 45760
    },
    {
      "epoch": 18.282461046743908,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0095,
      "step": 45761
    },
    {
      "epoch": 18.28286056731922,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 45762
    },
    {
      "epoch": 18.283260087894526,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 45763
    },
    {
      "epoch": 18.283659608469836,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 45764
    },
    {
      "epoch": 18.284059129045147,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 45765
    },
    {
      "epoch": 18.284458649620454,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 45766
    },
    {
      "epoch": 18.284858170195765,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 45767
    },
    {
      "epoch": 18.285257690771076,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0419,
      "step": 45768
    },
    {
      "epoch": 18.285657211346383,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 45769
    },
    {
      "epoch": 18.286056731921693,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 45770
    },
    {
      "epoch": 18.286456252497004,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 45771
    },
    {
      "epoch": 18.286855773072315,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 45772
    },
    {
      "epoch": 18.287255293647622,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 45773
    },
    {
      "epoch": 18.287654814222932,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 45774
    },
    {
      "epoch": 18.288054334798243,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 45775
    },
    {
      "epoch": 18.28845385537355,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 45776
    },
    {
      "epoch": 18.28885337594886,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 45777
    },
    {
      "epoch": 18.28925289652417,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 45778
    },
    {
      "epoch": 18.289652417099482,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 45779
    },
    {
      "epoch": 18.29005193767479,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 45780
    },
    {
      "epoch": 18.2904514582501,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 45781
    },
    {
      "epoch": 18.29085097882541,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 45782
    },
    {
      "epoch": 18.291250499400718,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 45783
    },
    {
      "epoch": 18.29165001997603,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 45784
    },
    {
      "epoch": 18.29204954055134,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 45785
    },
    {
      "epoch": 18.292449061126646,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 45786
    },
    {
      "epoch": 18.292848581701957,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 45787
    },
    {
      "epoch": 18.293248102277268,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 45788
    },
    {
      "epoch": 18.29364762285258,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 45789
    },
    {
      "epoch": 18.294047143427886,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 45790
    },
    {
      "epoch": 18.294446664003196,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0104,
      "step": 45791
    },
    {
      "epoch": 18.294846184578507,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 45792
    },
    {
      "epoch": 18.295245705153814,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 45793
    },
    {
      "epoch": 18.295645225729125,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 45794
    },
    {
      "epoch": 18.296044746304435,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 45795
    },
    {
      "epoch": 18.296444266879746,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 45796
    },
    {
      "epoch": 18.296843787455053,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 45797
    },
    {
      "epoch": 18.297243308030364,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 45798
    },
    {
      "epoch": 18.297642828605674,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 45799
    },
    {
      "epoch": 18.29804234918098,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 45800
    },
    {
      "epoch": 18.298441869756292,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 45801
    },
    {
      "epoch": 18.298841390331603,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 45802
    },
    {
      "epoch": 18.29924091090691,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 45803
    },
    {
      "epoch": 18.29964043148222,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 45804
    },
    {
      "epoch": 18.30003995205753,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 45805
    },
    {
      "epoch": 18.300439472632842,
      "grad_norm": 0.03759765625,
      "learning_rate": 0.0005,
      "loss": 1.0429,
      "step": 45806
    },
    {
      "epoch": 18.30083899320815,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 45807
    },
    {
      "epoch": 18.30123851378346,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 45808
    },
    {
      "epoch": 18.30163803435877,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 45809
    },
    {
      "epoch": 18.302037554934078,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 45810
    },
    {
      "epoch": 18.30243707550939,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 45811
    },
    {
      "epoch": 18.3028365960847,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 45812
    },
    {
      "epoch": 18.30323611666001,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 45813
    },
    {
      "epoch": 18.303635637235317,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 45814
    },
    {
      "epoch": 18.304035157810628,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 45815
    },
    {
      "epoch": 18.304434678385938,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 45816
    },
    {
      "epoch": 18.304834198961245,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 45817
    },
    {
      "epoch": 18.305233719536556,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 45818
    },
    {
      "epoch": 18.305633240111867,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 45819
    },
    {
      "epoch": 18.306032760687174,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 45820
    },
    {
      "epoch": 18.306432281262484,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 45821
    },
    {
      "epoch": 18.306831801837795,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 45822
    },
    {
      "epoch": 18.307231322413106,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 45823
    },
    {
      "epoch": 18.307630842988413,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 45824
    },
    {
      "epoch": 18.308030363563724,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 45825
    },
    {
      "epoch": 18.308429884139034,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 45826
    },
    {
      "epoch": 18.30882940471434,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 45827
    },
    {
      "epoch": 18.309228925289652,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 45828
    },
    {
      "epoch": 18.309628445864963,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 45829
    },
    {
      "epoch": 18.310027966440273,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0112,
      "step": 45830
    },
    {
      "epoch": 18.31042748701558,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 45831
    },
    {
      "epoch": 18.31082700759089,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 45832
    },
    {
      "epoch": 18.311226528166202,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0119,
      "step": 45833
    },
    {
      "epoch": 18.31162604874151,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 45834
    },
    {
      "epoch": 18.31202556931682,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 45835
    },
    {
      "epoch": 18.31242508989213,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 45836
    },
    {
      "epoch": 18.312824610467437,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 45837
    },
    {
      "epoch": 18.313224131042748,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 45838
    },
    {
      "epoch": 18.31362365161806,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 45839
    },
    {
      "epoch": 18.31402317219337,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 45840
    },
    {
      "epoch": 18.314422692768677,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 45841
    },
    {
      "epoch": 18.314822213343987,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 45842
    },
    {
      "epoch": 18.315221733919298,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 45843
    },
    {
      "epoch": 18.315621254494605,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 45844
    },
    {
      "epoch": 18.316020775069916,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 45845
    },
    {
      "epoch": 18.316420295645226,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 45846
    },
    {
      "epoch": 18.316819816220537,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 45847
    },
    {
      "epoch": 18.317219336795844,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 45848
    },
    {
      "epoch": 18.317618857371155,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 45849
    },
    {
      "epoch": 18.318018377946466,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 45850
    },
    {
      "epoch": 18.318417898521773,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 45851
    },
    {
      "epoch": 18.318817419097083,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 45852
    },
    {
      "epoch": 18.319216939672394,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 45853
    },
    {
      "epoch": 18.3196164602477,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 45854
    },
    {
      "epoch": 18.320015980823012,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 45855
    },
    {
      "epoch": 18.320415501398323,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 45856
    },
    {
      "epoch": 18.320815021973633,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 45857
    },
    {
      "epoch": 18.32121454254894,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 45858
    },
    {
      "epoch": 18.32161406312425,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 45859
    },
    {
      "epoch": 18.32201358369956,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 45860
    },
    {
      "epoch": 18.32241310427487,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 45861
    },
    {
      "epoch": 18.32281262485018,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 45862
    },
    {
      "epoch": 18.32321214542549,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 45863
    },
    {
      "epoch": 18.3236116660008,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 45864
    },
    {
      "epoch": 18.324011186576108,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 45865
    },
    {
      "epoch": 18.32441070715142,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 45866
    },
    {
      "epoch": 18.32481022772673,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 45867
    },
    {
      "epoch": 18.325209748302036,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 45868
    },
    {
      "epoch": 18.325609268877347,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 45869
    },
    {
      "epoch": 18.326008789452658,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 45870
    },
    {
      "epoch": 18.326408310027965,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 45871
    },
    {
      "epoch": 18.326807830603276,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 45872
    },
    {
      "epoch": 18.327207351178586,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 45873
    },
    {
      "epoch": 18.327606871753897,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 45874
    },
    {
      "epoch": 18.328006392329204,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 45875
    },
    {
      "epoch": 18.328405912904515,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 45876
    },
    {
      "epoch": 18.328805433479825,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 45877
    },
    {
      "epoch": 18.329204954055133,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 45878
    },
    {
      "epoch": 18.329604474630443,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 45879
    },
    {
      "epoch": 18.330003995205754,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 45880
    },
    {
      "epoch": 18.330403515781065,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 45881
    },
    {
      "epoch": 18.33080303635637,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 45882
    },
    {
      "epoch": 18.331202556931682,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 45883
    },
    {
      "epoch": 18.331602077506993,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 45884
    },
    {
      "epoch": 18.3320015980823,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 45885
    },
    {
      "epoch": 18.33240111865761,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 45886
    },
    {
      "epoch": 18.33280063923292,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0366,
      "step": 45887
    },
    {
      "epoch": 18.33320015980823,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 45888
    },
    {
      "epoch": 18.33359968038354,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 45889
    },
    {
      "epoch": 18.33399920095885,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 45890
    },
    {
      "epoch": 18.33439872153416,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 45891
    },
    {
      "epoch": 18.334798242109468,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 45892
    },
    {
      "epoch": 18.33519776268478,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 45893
    },
    {
      "epoch": 18.33559728326009,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 45894
    },
    {
      "epoch": 18.335996803835396,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 45895
    },
    {
      "epoch": 18.336396324410707,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 45896
    },
    {
      "epoch": 18.336795844986018,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 45897
    },
    {
      "epoch": 18.337195365561325,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 45898
    },
    {
      "epoch": 18.337594886136635,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 45899
    },
    {
      "epoch": 18.337994406711946,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 45900
    },
    {
      "epoch": 18.338393927287257,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 45901
    },
    {
      "epoch": 18.338793447862564,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 45902
    },
    {
      "epoch": 18.339192968437874,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 45903
    },
    {
      "epoch": 18.339592489013185,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 45904
    },
    {
      "epoch": 18.339992009588492,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.042,
      "step": 45905
    },
    {
      "epoch": 18.340391530163803,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 45906
    },
    {
      "epoch": 18.340791050739114,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 45907
    },
    {
      "epoch": 18.341190571314424,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 45908
    },
    {
      "epoch": 18.34159009188973,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 45909
    },
    {
      "epoch": 18.341989612465042,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 45910
    },
    {
      "epoch": 18.342389133040353,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 45911
    },
    {
      "epoch": 18.34278865361566,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 45912
    },
    {
      "epoch": 18.34318817419097,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 45913
    },
    {
      "epoch": 18.34358769476628,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 45914
    },
    {
      "epoch": 18.34398721534159,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 45915
    },
    {
      "epoch": 18.3443867359169,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 45916
    },
    {
      "epoch": 18.34478625649221,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 45917
    },
    {
      "epoch": 18.34518577706752,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 45918
    },
    {
      "epoch": 18.345585297642828,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 45919
    },
    {
      "epoch": 18.345984818218138,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 45920
    },
    {
      "epoch": 18.34638433879345,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 45921
    },
    {
      "epoch": 18.346783859368756,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 45922
    },
    {
      "epoch": 18.347183379944067,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 45923
    },
    {
      "epoch": 18.347582900519377,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 45924
    },
    {
      "epoch": 18.347982421094688,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 45925
    },
    {
      "epoch": 18.348381941669995,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 45926
    },
    {
      "epoch": 18.348781462245306,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 45927
    },
    {
      "epoch": 18.349180982820616,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 45928
    },
    {
      "epoch": 18.349580503395924,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0087,
      "step": 45929
    },
    {
      "epoch": 18.349980023971234,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 45930
    },
    {
      "epoch": 18.350379544546545,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 45931
    },
    {
      "epoch": 18.350779065121852,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 45932
    },
    {
      "epoch": 18.351178585697163,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 45933
    },
    {
      "epoch": 18.351578106272473,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 45934
    },
    {
      "epoch": 18.351977626847784,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 45935
    },
    {
      "epoch": 18.35237714742309,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 45936
    },
    {
      "epoch": 18.352776667998402,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 45937
    },
    {
      "epoch": 18.353176188573713,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 45938
    },
    {
      "epoch": 18.35357570914902,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 45939
    },
    {
      "epoch": 18.35397522972433,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0378,
      "step": 45940
    },
    {
      "epoch": 18.35437475029964,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 45941
    },
    {
      "epoch": 18.35477427087495,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 45942
    },
    {
      "epoch": 18.35517379145026,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0112,
      "step": 45943
    },
    {
      "epoch": 18.35557331202557,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 45944
    },
    {
      "epoch": 18.35597283260088,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 45945
    },
    {
      "epoch": 18.356372353176187,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 45946
    },
    {
      "epoch": 18.356771873751498,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 45947
    },
    {
      "epoch": 18.35717139432681,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 45948
    },
    {
      "epoch": 18.357570914902116,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 45949
    },
    {
      "epoch": 18.357970435477426,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 45950
    },
    {
      "epoch": 18.358369956052737,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 45951
    },
    {
      "epoch": 18.358769476628048,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 45952
    },
    {
      "epoch": 18.359168997203355,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 45953
    },
    {
      "epoch": 18.359568517778666,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 45954
    },
    {
      "epoch": 18.359968038353976,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 45955
    },
    {
      "epoch": 18.360367558929283,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 45956
    },
    {
      "epoch": 18.360767079504594,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 45957
    },
    {
      "epoch": 18.361166600079905,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 45958
    },
    {
      "epoch": 18.361566120655215,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 45959
    },
    {
      "epoch": 18.361965641230523,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 45960
    },
    {
      "epoch": 18.362365161805833,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 45961
    },
    {
      "epoch": 18.362764682381144,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 45962
    },
    {
      "epoch": 18.36316420295645,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 45963
    },
    {
      "epoch": 18.36356372353176,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 45964
    },
    {
      "epoch": 18.363963244107072,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 45965
    },
    {
      "epoch": 18.36436276468238,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 45966
    },
    {
      "epoch": 18.36476228525769,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 45967
    },
    {
      "epoch": 18.365161805833,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 45968
    },
    {
      "epoch": 18.36556132640831,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 45969
    },
    {
      "epoch": 18.36596084698362,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 45970
    },
    {
      "epoch": 18.36636036755893,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 45971
    },
    {
      "epoch": 18.36675988813424,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0116,
      "step": 45972
    },
    {
      "epoch": 18.367159408709547,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 45973
    },
    {
      "epoch": 18.367558929284858,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0126,
      "step": 45974
    },
    {
      "epoch": 18.36795844986017,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 45975
    },
    {
      "epoch": 18.36835797043548,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 45976
    },
    {
      "epoch": 18.368757491010786,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.003,
      "step": 45977
    },
    {
      "epoch": 18.369157011586097,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0098,
      "step": 45978
    },
    {
      "epoch": 18.369556532161408,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 45979
    },
    {
      "epoch": 18.369956052736715,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 45980
    },
    {
      "epoch": 18.370355573312025,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 45981
    },
    {
      "epoch": 18.370755093887336,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 45982
    },
    {
      "epoch": 18.371154614462643,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 45983
    },
    {
      "epoch": 18.371554135037954,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 45984
    },
    {
      "epoch": 18.371953655613265,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 45985
    },
    {
      "epoch": 18.372353176188575,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 45986
    },
    {
      "epoch": 18.372752696763882,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 45987
    },
    {
      "epoch": 18.373152217339193,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 45988
    },
    {
      "epoch": 18.373551737914504,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 45989
    },
    {
      "epoch": 18.37395125848981,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 45990
    },
    {
      "epoch": 18.37435077906512,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0073,
      "step": 45991
    },
    {
      "epoch": 18.374750299640432,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0066,
      "step": 45992
    },
    {
      "epoch": 18.375149820215743,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 45993
    },
    {
      "epoch": 18.37554934079105,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 45994
    },
    {
      "epoch": 18.37594886136636,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 45995
    },
    {
      "epoch": 18.37634838194167,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 45996
    },
    {
      "epoch": 18.37674790251698,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 45997
    },
    {
      "epoch": 18.37714742309229,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0117,
      "step": 45998
    },
    {
      "epoch": 18.3775469436676,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 45999
    },
    {
      "epoch": 18.377946464242907,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 46000
    },
    {
      "epoch": 18.378345984818218,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 46001
    },
    {
      "epoch": 18.37874550539353,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0373,
      "step": 46002
    },
    {
      "epoch": 18.37914502596884,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 46003
    },
    {
      "epoch": 18.379544546544146,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0449,
      "step": 46004
    },
    {
      "epoch": 18.379944067119457,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 46005
    },
    {
      "epoch": 18.380343587694767,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 46006
    },
    {
      "epoch": 18.380743108270075,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 46007
    },
    {
      "epoch": 18.381142628845385,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 46008
    },
    {
      "epoch": 18.381542149420696,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0124,
      "step": 46009
    },
    {
      "epoch": 18.381941669996007,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 46010
    },
    {
      "epoch": 18.382341190571314,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 46011
    },
    {
      "epoch": 18.382740711146624,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0361,
      "step": 46012
    },
    {
      "epoch": 18.383140231721935,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 46013
    },
    {
      "epoch": 18.383539752297242,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 46014
    },
    {
      "epoch": 18.383939272872553,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0105,
      "step": 46015
    },
    {
      "epoch": 18.384338793447863,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 46016
    },
    {
      "epoch": 18.38473831402317,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 46017
    },
    {
      "epoch": 18.38513783459848,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 46018
    },
    {
      "epoch": 18.385537355173792,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 46019
    },
    {
      "epoch": 18.385936875749103,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 46020
    },
    {
      "epoch": 18.38633639632441,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 46021
    },
    {
      "epoch": 18.38673591689972,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 46022
    },
    {
      "epoch": 18.38713543747503,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 46023
    },
    {
      "epoch": 18.387534958050338,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 46024
    },
    {
      "epoch": 18.38793447862565,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0064,
      "step": 46025
    },
    {
      "epoch": 18.38833399920096,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 46026
    },
    {
      "epoch": 18.388733519776267,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 46027
    },
    {
      "epoch": 18.389133040351577,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0069,
      "step": 46028
    },
    {
      "epoch": 18.389532560926888,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0067,
      "step": 46029
    },
    {
      "epoch": 18.3899320815022,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 46030
    },
    {
      "epoch": 18.390331602077506,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0072,
      "step": 46031
    },
    {
      "epoch": 18.390731122652817,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46032
    },
    {
      "epoch": 18.391130643228127,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 46033
    },
    {
      "epoch": 18.391530163803434,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 46034
    },
    {
      "epoch": 18.391929684378745,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 46035
    },
    {
      "epoch": 18.392329204954056,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0108,
      "step": 46036
    },
    {
      "epoch": 18.392728725529366,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46037
    },
    {
      "epoch": 18.393128246104673,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 46038
    },
    {
      "epoch": 18.393527766679984,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 46039
    },
    {
      "epoch": 18.393927287255295,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 46040
    },
    {
      "epoch": 18.394326807830602,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 46041
    },
    {
      "epoch": 18.394726328405913,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 46042
    },
    {
      "epoch": 18.395125848981223,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 46043
    },
    {
      "epoch": 18.39552536955653,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 46044
    },
    {
      "epoch": 18.39592489013184,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0342,
      "step": 46045
    },
    {
      "epoch": 18.39632441070715,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 46046
    },
    {
      "epoch": 18.396723931282462,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 46047
    },
    {
      "epoch": 18.39712345185777,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 46048
    },
    {
      "epoch": 18.39752297243308,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 46049
    },
    {
      "epoch": 18.39792249300839,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 46050
    },
    {
      "epoch": 18.398322013583698,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 46051
    },
    {
      "epoch": 18.39872153415901,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 46052
    },
    {
      "epoch": 18.39912105473432,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 46053
    },
    {
      "epoch": 18.39952057530963,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 46054
    },
    {
      "epoch": 18.399920095884937,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 46055
    },
    {
      "epoch": 18.400319616460248,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 46056
    },
    {
      "epoch": 18.40071913703556,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 46057
    },
    {
      "epoch": 18.401118657610866,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 46058
    },
    {
      "epoch": 18.401518178186176,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0432,
      "step": 46059
    },
    {
      "epoch": 18.401917698761487,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 46060
    },
    {
      "epoch": 18.402317219336794,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 46061
    },
    {
      "epoch": 18.402716739912105,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 46062
    },
    {
      "epoch": 18.403116260487415,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 46063
    },
    {
      "epoch": 18.403515781062726,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 46064
    },
    {
      "epoch": 18.403915301638033,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 46065
    },
    {
      "epoch": 18.404314822213344,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 46066
    },
    {
      "epoch": 18.404714342788655,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 46067
    },
    {
      "epoch": 18.40511386336396,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 46068
    },
    {
      "epoch": 18.405513383939272,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 46069
    },
    {
      "epoch": 18.405912904514583,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 46070
    },
    {
      "epoch": 18.406312425089894,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46071
    },
    {
      "epoch": 18.4067119456652,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 46072
    },
    {
      "epoch": 18.40711146624051,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 46073
    },
    {
      "epoch": 18.407510986815822,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 46074
    },
    {
      "epoch": 18.40791050739113,
      "grad_norm": 0.0380859375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 46075
    },
    {
      "epoch": 18.40831002796644,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 46076
    },
    {
      "epoch": 18.40870954854175,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 46077
    },
    {
      "epoch": 18.409109069117058,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 46078
    },
    {
      "epoch": 18.40950858969237,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 46079
    },
    {
      "epoch": 18.40990811026768,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 46080
    },
    {
      "epoch": 18.41030763084299,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 46081
    },
    {
      "epoch": 18.410707151418297,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 46082
    },
    {
      "epoch": 18.411106671993608,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 46083
    },
    {
      "epoch": 18.41150619256892,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 46084
    },
    {
      "epoch": 18.411905713144225,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 46085
    },
    {
      "epoch": 18.412305233719536,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 46086
    },
    {
      "epoch": 18.412704754294847,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 46087
    },
    {
      "epoch": 18.413104274870157,
      "grad_norm": 0.036865234375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 46088
    },
    {
      "epoch": 18.413503795445465,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 46089
    },
    {
      "epoch": 18.413903316020775,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 46090
    },
    {
      "epoch": 18.414302836596086,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 46091
    },
    {
      "epoch": 18.414702357171393,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 46092
    },
    {
      "epoch": 18.415101877746704,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 46093
    },
    {
      "epoch": 18.415501398322014,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 46094
    },
    {
      "epoch": 18.41590091889732,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 46095
    },
    {
      "epoch": 18.416300439472632,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46096
    },
    {
      "epoch": 18.416699960047943,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0426,
      "step": 46097
    },
    {
      "epoch": 18.417099480623254,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 46098
    },
    {
      "epoch": 18.41749900119856,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 46099
    },
    {
      "epoch": 18.41789852177387,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 46100
    },
    {
      "epoch": 18.418298042349182,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 46101
    },
    {
      "epoch": 18.41869756292449,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 46102
    },
    {
      "epoch": 18.4190970834998,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 46103
    },
    {
      "epoch": 18.41949660407511,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 46104
    },
    {
      "epoch": 18.41989612465042,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 46105
    },
    {
      "epoch": 18.42029564522573,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46106
    },
    {
      "epoch": 18.42069516580104,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 46107
    },
    {
      "epoch": 18.42109468637635,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 46108
    },
    {
      "epoch": 18.421494206951657,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 46109
    },
    {
      "epoch": 18.421893727526967,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 46110
    },
    {
      "epoch": 18.422293248102278,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 46111
    },
    {
      "epoch": 18.422692768677585,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 46112
    },
    {
      "epoch": 18.423092289252896,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 46113
    },
    {
      "epoch": 18.423491809828207,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 46114
    },
    {
      "epoch": 18.423891330403517,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0094,
      "step": 46115
    },
    {
      "epoch": 18.424290850978824,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46116
    },
    {
      "epoch": 18.424690371554135,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 46117
    },
    {
      "epoch": 18.425089892129446,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 46118
    },
    {
      "epoch": 18.425489412704753,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 46119
    },
    {
      "epoch": 18.425888933280063,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46120
    },
    {
      "epoch": 18.426288453855374,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 46121
    },
    {
      "epoch": 18.426687974430685,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 46122
    },
    {
      "epoch": 18.427087495005992,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 46123
    },
    {
      "epoch": 18.427487015581303,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 46124
    },
    {
      "epoch": 18.427886536156613,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 46125
    },
    {
      "epoch": 18.42828605673192,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 46126
    },
    {
      "epoch": 18.42868557730723,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 46127
    },
    {
      "epoch": 18.429085097882542,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 46128
    },
    {
      "epoch": 18.42948461845785,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 46129
    },
    {
      "epoch": 18.42988413903316,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 46130
    },
    {
      "epoch": 18.43028365960847,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 46131
    },
    {
      "epoch": 18.43068318018378,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 46132
    },
    {
      "epoch": 18.431082700759088,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 46133
    },
    {
      "epoch": 18.4314822213344,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 46134
    },
    {
      "epoch": 18.43188174190971,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 46135
    },
    {
      "epoch": 18.432281262485017,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 46136
    },
    {
      "epoch": 18.432680783060327,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 46137
    },
    {
      "epoch": 18.433080303635638,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 46138
    },
    {
      "epoch": 18.43347982421095,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 46139
    },
    {
      "epoch": 18.433879344786256,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 46140
    },
    {
      "epoch": 18.434278865361566,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 46141
    },
    {
      "epoch": 18.434678385936877,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 46142
    },
    {
      "epoch": 18.435077906512184,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 46143
    },
    {
      "epoch": 18.435477427087495,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 46144
    },
    {
      "epoch": 18.435876947662805,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 46145
    },
    {
      "epoch": 18.436276468238113,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 46146
    },
    {
      "epoch": 18.436675988813423,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 46147
    },
    {
      "epoch": 18.437075509388734,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 46148
    },
    {
      "epoch": 18.437475029964045,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 46149
    },
    {
      "epoch": 18.43787455053935,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0005,
      "loss": 1.0398,
      "step": 46150
    },
    {
      "epoch": 18.438274071114662,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 46151
    },
    {
      "epoch": 18.438673591689973,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 46152
    },
    {
      "epoch": 18.43907311226528,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 46153
    },
    {
      "epoch": 18.43947263284059,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 46154
    },
    {
      "epoch": 18.4398721534159,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 46155
    },
    {
      "epoch": 18.440271673991212,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 46156
    },
    {
      "epoch": 18.44067119456652,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 46157
    },
    {
      "epoch": 18.44107071514183,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 46158
    },
    {
      "epoch": 18.44147023571714,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 46159
    },
    {
      "epoch": 18.441869756292448,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 46160
    },
    {
      "epoch": 18.44226927686776,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 46161
    },
    {
      "epoch": 18.44266879744307,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 46162
    },
    {
      "epoch": 18.443068318018376,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 46163
    },
    {
      "epoch": 18.443467838593687,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0163,
      "step": 46164
    },
    {
      "epoch": 18.443867359168998,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 46165
    },
    {
      "epoch": 18.44426687974431,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 46166
    },
    {
      "epoch": 18.444666400319615,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 46167
    },
    {
      "epoch": 18.445065920894926,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0108,
      "step": 46168
    },
    {
      "epoch": 18.445465441470237,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 46169
    },
    {
      "epoch": 18.445864962045544,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 46170
    },
    {
      "epoch": 18.446264482620855,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 46171
    },
    {
      "epoch": 18.446664003196165,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 46172
    },
    {
      "epoch": 18.447063523771476,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 46173
    },
    {
      "epoch": 18.447463044346783,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 46174
    },
    {
      "epoch": 18.447862564922094,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 46175
    },
    {
      "epoch": 18.448262085497404,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 46176
    },
    {
      "epoch": 18.44866160607271,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 46177
    },
    {
      "epoch": 18.449061126648022,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 46178
    },
    {
      "epoch": 18.449460647223333,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 46179
    },
    {
      "epoch": 18.44986016779864,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0106,
      "step": 46180
    },
    {
      "epoch": 18.45025968837395,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0408,
      "step": 46181
    },
    {
      "epoch": 18.45065920894926,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 46182
    },
    {
      "epoch": 18.451058729524572,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 46183
    },
    {
      "epoch": 18.45145825009988,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 46184
    },
    {
      "epoch": 18.45185777067519,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 46185
    },
    {
      "epoch": 18.4522572912505,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 46186
    },
    {
      "epoch": 18.452656811825808,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 46187
    },
    {
      "epoch": 18.45305633240112,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 46188
    },
    {
      "epoch": 18.45345585297643,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 46189
    },
    {
      "epoch": 18.45385537355174,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 46190
    },
    {
      "epoch": 18.454254894127047,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46191
    },
    {
      "epoch": 18.454654414702357,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 46192
    },
    {
      "epoch": 18.455053935277668,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 46193
    },
    {
      "epoch": 18.455453455852975,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 46194
    },
    {
      "epoch": 18.455852976428286,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 46195
    },
    {
      "epoch": 18.456252497003597,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 46196
    },
    {
      "epoch": 18.456652017578904,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 46197
    },
    {
      "epoch": 18.457051538154214,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 46198
    },
    {
      "epoch": 18.457451058729525,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 46199
    },
    {
      "epoch": 18.457850579304836,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.012,
      "step": 46200
    },
    {
      "epoch": 18.458250099880143,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 46201
    },
    {
      "epoch": 18.458649620455454,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 46202
    },
    {
      "epoch": 18.459049141030764,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 46203
    },
    {
      "epoch": 18.45944866160607,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 46204
    },
    {
      "epoch": 18.459848182181382,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 46205
    },
    {
      "epoch": 18.460247702756693,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 46206
    },
    {
      "epoch": 18.460647223332003,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 46207
    },
    {
      "epoch": 18.46104674390731,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 46208
    },
    {
      "epoch": 18.46144626448262,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 46209
    },
    {
      "epoch": 18.461845785057932,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 46210
    },
    {
      "epoch": 18.46224530563324,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 46211
    },
    {
      "epoch": 18.46264482620855,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 46212
    },
    {
      "epoch": 18.46304434678386,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 46213
    },
    {
      "epoch": 18.463443867359167,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 46214
    },
    {
      "epoch": 18.463843387934478,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 46215
    },
    {
      "epoch": 18.46424290850979,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0119,
      "step": 46216
    },
    {
      "epoch": 18.4646424290851,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 46217
    },
    {
      "epoch": 18.465041949660407,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 46218
    },
    {
      "epoch": 18.465441470235717,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 46219
    },
    {
      "epoch": 18.465840990811028,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 46220
    },
    {
      "epoch": 18.466240511386335,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 46221
    },
    {
      "epoch": 18.466640031961646,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 46222
    },
    {
      "epoch": 18.467039552536956,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 46223
    },
    {
      "epoch": 18.467439073112267,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 46224
    },
    {
      "epoch": 18.467838593687574,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 46225
    },
    {
      "epoch": 18.468238114262885,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 46226
    },
    {
      "epoch": 18.468637634838196,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 46227
    },
    {
      "epoch": 18.469037155413503,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 46228
    },
    {
      "epoch": 18.469436675988813,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 46229
    },
    {
      "epoch": 18.469836196564124,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 46230
    },
    {
      "epoch": 18.47023571713943,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 46231
    },
    {
      "epoch": 18.470635237714742,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 46232
    },
    {
      "epoch": 18.471034758290052,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 46233
    },
    {
      "epoch": 18.471434278865363,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 46234
    },
    {
      "epoch": 18.47183379944067,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 46235
    },
    {
      "epoch": 18.47223332001598,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 46236
    },
    {
      "epoch": 18.47263284059129,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 46237
    },
    {
      "epoch": 18.4730323611666,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 46238
    },
    {
      "epoch": 18.47343188174191,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46239
    },
    {
      "epoch": 18.47383140231722,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 46240
    },
    {
      "epoch": 18.474230922892527,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 46241
    },
    {
      "epoch": 18.474630443467838,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 46242
    },
    {
      "epoch": 18.47502996404315,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 46243
    },
    {
      "epoch": 18.47542948461846,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 46244
    },
    {
      "epoch": 18.475829005193766,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 46245
    },
    {
      "epoch": 18.476228525769077,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 46246
    },
    {
      "epoch": 18.476628046344388,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 46247
    },
    {
      "epoch": 18.477027566919695,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 46248
    },
    {
      "epoch": 18.477427087495006,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0058,
      "step": 46249
    },
    {
      "epoch": 18.477826608070316,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 46250
    },
    {
      "epoch": 18.478226128645627,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 46251
    },
    {
      "epoch": 18.478625649220934,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0101,
      "step": 46252
    },
    {
      "epoch": 18.479025169796245,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 46253
    },
    {
      "epoch": 18.479424690371555,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 46254
    },
    {
      "epoch": 18.479824210946862,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 46255
    },
    {
      "epoch": 18.480223731522173,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 46256
    },
    {
      "epoch": 18.480623252097484,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 46257
    },
    {
      "epoch": 18.48102277267279,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 46258
    },
    {
      "epoch": 18.4814222932481,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 46259
    },
    {
      "epoch": 18.481821813823412,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 46260
    },
    {
      "epoch": 18.482221334398723,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 46261
    },
    {
      "epoch": 18.48262085497403,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 46262
    },
    {
      "epoch": 18.48302037554934,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 46263
    },
    {
      "epoch": 18.48341989612465,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 46264
    },
    {
      "epoch": 18.48381941669996,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 46265
    },
    {
      "epoch": 18.48421893727527,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 46266
    },
    {
      "epoch": 18.48461845785058,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 46267
    },
    {
      "epoch": 18.48501797842589,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 46268
    },
    {
      "epoch": 18.485417499001198,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 46269
    },
    {
      "epoch": 18.48581701957651,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 46270
    },
    {
      "epoch": 18.48621654015182,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 46271
    },
    {
      "epoch": 18.486616060727126,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 46272
    },
    {
      "epoch": 18.487015581302437,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 46273
    },
    {
      "epoch": 18.487415101877748,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0107,
      "step": 46274
    },
    {
      "epoch": 18.487814622453055,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 46275
    },
    {
      "epoch": 18.488214143028365,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 46276
    },
    {
      "epoch": 18.488613663603676,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 46277
    },
    {
      "epoch": 18.489013184178987,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 46278
    },
    {
      "epoch": 18.489412704754294,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 46279
    },
    {
      "epoch": 18.489812225329604,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 46280
    },
    {
      "epoch": 18.490211745904915,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 46281
    },
    {
      "epoch": 18.490611266480222,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 46282
    },
    {
      "epoch": 18.491010787055533,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 46283
    },
    {
      "epoch": 18.491410307630844,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 46284
    },
    {
      "epoch": 18.491809828206154,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 46285
    },
    {
      "epoch": 18.49220934878146,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 46286
    },
    {
      "epoch": 18.492608869356772,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 46287
    },
    {
      "epoch": 18.493008389932083,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 46288
    },
    {
      "epoch": 18.49340791050739,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 46289
    },
    {
      "epoch": 18.4938074310827,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 46290
    },
    {
      "epoch": 18.49420695165801,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 46291
    },
    {
      "epoch": 18.49460647223332,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 46292
    },
    {
      "epoch": 18.49500599280863,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 46293
    },
    {
      "epoch": 18.49540551338394,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 46294
    },
    {
      "epoch": 18.49580503395925,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 46295
    },
    {
      "epoch": 18.496204554534557,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 46296
    },
    {
      "epoch": 18.496604075109868,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 46297
    },
    {
      "epoch": 18.49700359568518,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 46298
    },
    {
      "epoch": 18.497403116260486,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 46299
    },
    {
      "epoch": 18.497802636835797,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 46300
    },
    {
      "epoch": 18.498202157411107,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 46301
    },
    {
      "epoch": 18.498601677986418,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 46302
    },
    {
      "epoch": 18.499001198561725,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 46303
    },
    {
      "epoch": 18.499400719137036,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 46304
    },
    {
      "epoch": 18.499800239712346,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 46305
    },
    {
      "epoch": 18.500199760287654,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 46306
    },
    {
      "epoch": 18.500599280862964,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 46307
    },
    {
      "epoch": 18.500998801438275,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 46308
    },
    {
      "epoch": 18.501398322013582,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 46309
    },
    {
      "epoch": 18.501797842588893,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 46310
    },
    {
      "epoch": 18.502197363164203,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 46311
    },
    {
      "epoch": 18.502596883739514,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 46312
    },
    {
      "epoch": 18.50299640431482,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 46313
    },
    {
      "epoch": 18.503395924890132,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 46314
    },
    {
      "epoch": 18.503795445465443,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 46315
    },
    {
      "epoch": 18.50419496604075,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 46316
    },
    {
      "epoch": 18.50459448661606,
      "grad_norm": 0.1875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 46317
    },
    {
      "epoch": 18.50499400719137,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 46318
    },
    {
      "epoch": 18.50539352776668,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 46319
    },
    {
      "epoch": 18.50579304834199,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 46320
    },
    {
      "epoch": 18.5061925689173,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0124,
      "step": 46321
    },
    {
      "epoch": 18.50659208949261,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 46322
    },
    {
      "epoch": 18.506991610067917,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 46323
    },
    {
      "epoch": 18.507391130643228,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 46324
    },
    {
      "epoch": 18.50779065121854,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 46325
    },
    {
      "epoch": 18.508190171793846,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 46326
    },
    {
      "epoch": 18.508589692369156,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 46327
    },
    {
      "epoch": 18.508989212944467,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 46328
    },
    {
      "epoch": 18.509388733519778,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 46329
    },
    {
      "epoch": 18.509788254095085,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 46330
    },
    {
      "epoch": 18.510187774670396,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 46331
    },
    {
      "epoch": 18.510587295245706,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 46332
    },
    {
      "epoch": 18.510986815821013,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 46333
    },
    {
      "epoch": 18.511386336396324,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 46334
    },
    {
      "epoch": 18.511785856971635,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 46335
    },
    {
      "epoch": 18.512185377546945,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 46336
    },
    {
      "epoch": 18.512584898122252,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 46337
    },
    {
      "epoch": 18.512984418697563,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 46338
    },
    {
      "epoch": 18.513383939272874,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 46339
    },
    {
      "epoch": 18.51378345984818,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 46340
    },
    {
      "epoch": 18.51418298042349,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 46341
    },
    {
      "epoch": 18.514582500998802,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 46342
    },
    {
      "epoch": 18.51498202157411,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 46343
    },
    {
      "epoch": 18.51538154214942,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 46344
    },
    {
      "epoch": 18.51578106272473,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 46345
    },
    {
      "epoch": 18.51618058330004,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 46346
    },
    {
      "epoch": 18.51658010387535,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 46347
    },
    {
      "epoch": 18.51697962445066,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 46348
    },
    {
      "epoch": 18.51737914502597,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 46349
    },
    {
      "epoch": 18.517778665601277,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 46350
    },
    {
      "epoch": 18.518178186176588,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0349,
      "step": 46351
    },
    {
      "epoch": 18.5185777067519,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 46352
    },
    {
      "epoch": 18.518977227327206,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 46353
    },
    {
      "epoch": 18.519376747902516,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 46354
    },
    {
      "epoch": 18.519776268477827,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 46355
    },
    {
      "epoch": 18.520175789053138,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 46356
    },
    {
      "epoch": 18.520575309628445,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 46357
    },
    {
      "epoch": 18.520974830203755,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 46358
    },
    {
      "epoch": 18.521374350779066,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 46359
    },
    {
      "epoch": 18.521773871354373,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 46360
    },
    {
      "epoch": 18.522173391929684,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 46361
    },
    {
      "epoch": 18.522572912504994,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0108,
      "step": 46362
    },
    {
      "epoch": 18.522972433080305,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 46363
    },
    {
      "epoch": 18.523371953655612,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 46364
    },
    {
      "epoch": 18.523771474230923,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 46365
    },
    {
      "epoch": 18.524170994806234,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 46366
    },
    {
      "epoch": 18.52457051538154,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 46367
    },
    {
      "epoch": 18.52497003595685,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 46368
    },
    {
      "epoch": 18.525369556532162,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 46369
    },
    {
      "epoch": 18.52576907710747,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 46370
    },
    {
      "epoch": 18.52616859768278,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 46371
    },
    {
      "epoch": 18.52656811825809,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 46372
    },
    {
      "epoch": 18.5269676388334,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 46373
    },
    {
      "epoch": 18.52736715940871,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0065,
      "step": 46374
    },
    {
      "epoch": 18.52776667998402,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 46375
    },
    {
      "epoch": 18.52816620055933,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 46376
    },
    {
      "epoch": 18.528565721134637,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 46377
    },
    {
      "epoch": 18.528965241709948,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 46378
    },
    {
      "epoch": 18.529364762285258,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 46379
    },
    {
      "epoch": 18.52976428286057,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 46380
    },
    {
      "epoch": 18.530163803435876,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 46381
    },
    {
      "epoch": 18.530563324011187,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 46382
    },
    {
      "epoch": 18.530962844586497,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 46383
    },
    {
      "epoch": 18.531362365161804,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 46384
    },
    {
      "epoch": 18.531761885737115,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 46385
    },
    {
      "epoch": 18.532161406312426,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 46386
    },
    {
      "epoch": 18.532560926887733,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 46387
    },
    {
      "epoch": 18.532960447463044,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 46388
    },
    {
      "epoch": 18.533359968038354,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0124,
      "step": 46389
    },
    {
      "epoch": 18.533759488613665,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 46390
    },
    {
      "epoch": 18.534159009188972,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 46391
    },
    {
      "epoch": 18.534558529764283,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46392
    },
    {
      "epoch": 18.534958050339593,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 46393
    },
    {
      "epoch": 18.5353575709149,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 46394
    },
    {
      "epoch": 18.53575709149021,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46395
    },
    {
      "epoch": 18.536156612065522,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 46396
    },
    {
      "epoch": 18.536556132640833,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 46397
    },
    {
      "epoch": 18.53695565321614,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 46398
    },
    {
      "epoch": 18.53735517379145,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 46399
    },
    {
      "epoch": 18.53775469436676,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 46400
    },
    {
      "epoch": 18.538154214942068,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 46401
    },
    {
      "epoch": 18.53855373551738,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 46402
    },
    {
      "epoch": 18.53895325609269,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 46403
    },
    {
      "epoch": 18.539352776667997,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 46404
    },
    {
      "epoch": 18.539752297243307,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 46405
    },
    {
      "epoch": 18.540151817818618,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 46406
    },
    {
      "epoch": 18.54055133839393,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 46407
    },
    {
      "epoch": 18.540950858969236,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 46408
    },
    {
      "epoch": 18.541350379544546,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 46409
    },
    {
      "epoch": 18.541749900119857,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 46410
    },
    {
      "epoch": 18.542149420695164,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 46411
    },
    {
      "epoch": 18.542548941270475,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 46412
    },
    {
      "epoch": 18.542948461845786,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 46413
    },
    {
      "epoch": 18.543347982421096,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 46414
    },
    {
      "epoch": 18.543747502996403,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 46415
    },
    {
      "epoch": 18.544147023571714,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 46416
    },
    {
      "epoch": 18.544546544147025,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 46417
    },
    {
      "epoch": 18.544946064722332,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 46418
    },
    {
      "epoch": 18.545345585297643,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 46419
    },
    {
      "epoch": 18.545745105872953,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 46420
    },
    {
      "epoch": 18.54614462644826,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 46421
    },
    {
      "epoch": 18.54654414702357,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0054,
      "step": 46422
    },
    {
      "epoch": 18.54694366759888,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 46423
    },
    {
      "epoch": 18.547343188174192,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 46424
    },
    {
      "epoch": 18.5477427087495,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 46425
    },
    {
      "epoch": 18.54814222932481,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 46426
    },
    {
      "epoch": 18.54854174990012,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 46427
    },
    {
      "epoch": 18.548941270475428,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 46428
    },
    {
      "epoch": 18.54934079105074,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 46429
    },
    {
      "epoch": 18.54974031162605,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 46430
    },
    {
      "epoch": 18.55013983220136,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0122,
      "step": 46431
    },
    {
      "epoch": 18.550539352776667,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 46432
    },
    {
      "epoch": 18.550938873351978,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 46433
    },
    {
      "epoch": 18.55133839392729,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 46434
    },
    {
      "epoch": 18.551737914502596,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 46435
    },
    {
      "epoch": 18.552137435077906,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 46436
    },
    {
      "epoch": 18.552536955653217,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 46437
    },
    {
      "epoch": 18.552936476228524,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0113,
      "step": 46438
    },
    {
      "epoch": 18.553335996803835,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 46439
    },
    {
      "epoch": 18.553735517379145,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 46440
    },
    {
      "epoch": 18.554135037954456,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 46441
    },
    {
      "epoch": 18.554534558529763,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 46442
    },
    {
      "epoch": 18.554934079105074,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 46443
    },
    {
      "epoch": 18.555333599680385,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 46444
    },
    {
      "epoch": 18.55573312025569,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 46445
    },
    {
      "epoch": 18.556132640831002,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 46446
    },
    {
      "epoch": 18.556532161406313,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 46447
    },
    {
      "epoch": 18.556931681981624,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 46448
    },
    {
      "epoch": 18.55733120255693,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 46449
    },
    {
      "epoch": 18.55773072313224,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 46450
    },
    {
      "epoch": 18.558130243707552,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 46451
    },
    {
      "epoch": 18.55852976428286,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 46452
    },
    {
      "epoch": 18.55892928485817,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 46453
    },
    {
      "epoch": 18.55932880543348,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 46454
    },
    {
      "epoch": 18.559728326008788,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 46455
    },
    {
      "epoch": 18.5601278465841,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 46456
    },
    {
      "epoch": 18.56052736715941,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 46457
    },
    {
      "epoch": 18.56092688773472,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 46458
    },
    {
      "epoch": 18.561326408310027,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 46459
    },
    {
      "epoch": 18.561725928885338,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 46460
    },
    {
      "epoch": 18.56212544946065,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 46461
    },
    {
      "epoch": 18.562524970035955,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 46462
    },
    {
      "epoch": 18.562924490611266,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 46463
    },
    {
      "epoch": 18.563324011186577,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0056,
      "step": 46464
    },
    {
      "epoch": 18.563723531761887,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 46465
    },
    {
      "epoch": 18.564123052337195,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 46466
    },
    {
      "epoch": 18.564522572912505,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0369,
      "step": 46467
    },
    {
      "epoch": 18.564922093487816,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 46468
    },
    {
      "epoch": 18.565321614063123,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 46469
    },
    {
      "epoch": 18.565721134638434,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0005,
      "loss": 1.0353,
      "step": 46470
    },
    {
      "epoch": 18.566120655213744,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 46471
    },
    {
      "epoch": 18.56652017578905,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 46472
    },
    {
      "epoch": 18.566919696364362,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 46473
    },
    {
      "epoch": 18.567319216939673,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 46474
    },
    {
      "epoch": 18.567718737514983,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0005,
      "loss": 1.0456,
      "step": 46475
    },
    {
      "epoch": 18.56811825809029,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 46476
    },
    {
      "epoch": 18.5685177786656,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 46477
    },
    {
      "epoch": 18.568917299240912,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 46478
    },
    {
      "epoch": 18.56931681981622,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 46479
    },
    {
      "epoch": 18.56971634039153,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 46480
    },
    {
      "epoch": 18.57011586096684,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 46481
    },
    {
      "epoch": 18.57051538154215,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 46482
    },
    {
      "epoch": 18.570914902117458,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 46483
    },
    {
      "epoch": 18.57131442269277,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 46484
    },
    {
      "epoch": 18.57171394326808,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 46485
    },
    {
      "epoch": 18.572113463843387,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 46486
    },
    {
      "epoch": 18.572512984418697,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 46487
    },
    {
      "epoch": 18.572912504994008,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 46488
    },
    {
      "epoch": 18.573312025569315,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 46489
    },
    {
      "epoch": 18.573711546144626,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 46490
    },
    {
      "epoch": 18.574111066719937,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 46491
    },
    {
      "epoch": 18.574510587295247,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 46492
    },
    {
      "epoch": 18.574910107870554,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0135,
      "step": 46493
    },
    {
      "epoch": 18.575309628445865,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 46494
    },
    {
      "epoch": 18.575709149021176,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 46495
    },
    {
      "epoch": 18.576108669596483,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 46496
    },
    {
      "epoch": 18.576508190171793,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 46497
    },
    {
      "epoch": 18.576907710747104,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 46498
    },
    {
      "epoch": 18.577307231322415,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 46499
    },
    {
      "epoch": 18.577706751897722,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 46500
    },
    {
      "epoch": 18.578106272473033,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 46501
    },
    {
      "epoch": 18.578505793048343,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 46502
    },
    {
      "epoch": 18.57890531362365,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 46503
    },
    {
      "epoch": 18.57930483419896,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 46504
    },
    {
      "epoch": 18.57970435477427,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 46505
    },
    {
      "epoch": 18.58010387534958,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 46506
    },
    {
      "epoch": 18.58050339592489,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 46507
    },
    {
      "epoch": 18.5809029165002,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 46508
    },
    {
      "epoch": 18.58130243707551,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 46509
    },
    {
      "epoch": 18.581701957650818,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 46510
    },
    {
      "epoch": 18.58210147822613,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 46511
    },
    {
      "epoch": 18.58250099880144,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 46512
    },
    {
      "epoch": 18.582900519376746,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 46513
    },
    {
      "epoch": 18.583300039952057,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 46514
    },
    {
      "epoch": 18.583699560527368,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 46515
    },
    {
      "epoch": 18.58409908110268,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46516
    },
    {
      "epoch": 18.584498601677986,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 46517
    },
    {
      "epoch": 18.584898122253296,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 46518
    },
    {
      "epoch": 18.585297642828607,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 46519
    },
    {
      "epoch": 18.585697163403914,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 46520
    },
    {
      "epoch": 18.586096683979225,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 46521
    },
    {
      "epoch": 18.586496204554535,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 46522
    },
    {
      "epoch": 18.586895725129843,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 46523
    },
    {
      "epoch": 18.587295245705153,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 46524
    },
    {
      "epoch": 18.587694766280464,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 46525
    },
    {
      "epoch": 18.588094286855775,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 46526
    },
    {
      "epoch": 18.58849380743108,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 46527
    },
    {
      "epoch": 18.588893328006392,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 46528
    },
    {
      "epoch": 18.589292848581703,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 46529
    },
    {
      "epoch": 18.58969236915701,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 46530
    },
    {
      "epoch": 18.59009188973232,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 46531
    },
    {
      "epoch": 18.59049141030763,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0403,
      "step": 46532
    },
    {
      "epoch": 18.590890930882942,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 46533
    },
    {
      "epoch": 18.59129045145825,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 46534
    },
    {
      "epoch": 18.59168997203356,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0135,
      "step": 46535
    },
    {
      "epoch": 18.59208949260887,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 46536
    },
    {
      "epoch": 18.592489013184178,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 46537
    },
    {
      "epoch": 18.59288853375949,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 46538
    },
    {
      "epoch": 18.5932880543348,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 46539
    },
    {
      "epoch": 18.593687574910106,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 46540
    },
    {
      "epoch": 18.594087095485417,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 46541
    },
    {
      "epoch": 18.594486616060728,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 46542
    },
    {
      "epoch": 18.59488613663604,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 46543
    },
    {
      "epoch": 18.595285657211345,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 46544
    },
    {
      "epoch": 18.595685177786656,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 46545
    },
    {
      "epoch": 18.596084698361967,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 46546
    },
    {
      "epoch": 18.596484218937274,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 46547
    },
    {
      "epoch": 18.596883739512585,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 46548
    },
    {
      "epoch": 18.597283260087895,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0121,
      "step": 46549
    },
    {
      "epoch": 18.597682780663206,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 46550
    },
    {
      "epoch": 18.598082301238513,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 46551
    },
    {
      "epoch": 18.598481821813824,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 46552
    },
    {
      "epoch": 18.598881342389134,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 46553
    },
    {
      "epoch": 18.59928086296444,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 46554
    },
    {
      "epoch": 18.599680383539752,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 46555
    },
    {
      "epoch": 18.600079904115063,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 46556
    },
    {
      "epoch": 18.60047942469037,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 46557
    },
    {
      "epoch": 18.60087894526568,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 46558
    },
    {
      "epoch": 18.60127846584099,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 46559
    },
    {
      "epoch": 18.601677986416302,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 46560
    },
    {
      "epoch": 18.60207750699161,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 46561
    },
    {
      "epoch": 18.60247702756692,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 46562
    },
    {
      "epoch": 18.60287654814223,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 46563
    },
    {
      "epoch": 18.603276068717538,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 46564
    },
    {
      "epoch": 18.60367558929285,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 46565
    },
    {
      "epoch": 18.60407510986816,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 46566
    },
    {
      "epoch": 18.60447463044347,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 46567
    },
    {
      "epoch": 18.604874151018777,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 46568
    },
    {
      "epoch": 18.605273671594087,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.0093,
      "step": 46569
    },
    {
      "epoch": 18.605673192169398,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 46570
    },
    {
      "epoch": 18.606072712744705,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 46571
    },
    {
      "epoch": 18.606472233320016,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 46572
    },
    {
      "epoch": 18.606871753895327,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 46573
    },
    {
      "epoch": 18.607271274470634,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 46574
    },
    {
      "epoch": 18.607670795045944,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 46575
    },
    {
      "epoch": 18.608070315621255,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 46576
    },
    {
      "epoch": 18.608469836196566,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 46577
    },
    {
      "epoch": 18.608869356771873,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 46578
    },
    {
      "epoch": 18.609268877347183,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 46579
    },
    {
      "epoch": 18.609668397922494,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 46580
    },
    {
      "epoch": 18.6100679184978,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 46581
    },
    {
      "epoch": 18.610467439073112,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 46582
    },
    {
      "epoch": 18.610866959648423,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0131,
      "step": 46583
    },
    {
      "epoch": 18.611266480223733,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 46584
    },
    {
      "epoch": 18.61166600079904,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 46585
    },
    {
      "epoch": 18.61206552137435,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 46586
    },
    {
      "epoch": 18.612465041949662,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 46587
    },
    {
      "epoch": 18.61286456252497,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 46588
    },
    {
      "epoch": 18.61326408310028,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 46589
    },
    {
      "epoch": 18.61366360367559,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 46590
    },
    {
      "epoch": 18.614063124250897,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 46591
    },
    {
      "epoch": 18.614462644826208,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 46592
    },
    {
      "epoch": 18.61486216540152,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 46593
    },
    {
      "epoch": 18.61526168597683,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 46594
    },
    {
      "epoch": 18.615661206552137,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 46595
    },
    {
      "epoch": 18.616060727127447,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 46596
    },
    {
      "epoch": 18.616460247702758,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 46597
    },
    {
      "epoch": 18.616859768278065,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 46598
    },
    {
      "epoch": 18.617259288853376,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 46599
    },
    {
      "epoch": 18.617658809428686,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 46600
    },
    {
      "epoch": 18.618058330003997,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 46601
    },
    {
      "epoch": 18.618457850579304,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 46602
    },
    {
      "epoch": 18.618857371154615,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 46603
    },
    {
      "epoch": 18.619256891729925,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 46604
    },
    {
      "epoch": 18.619656412305233,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 46605
    },
    {
      "epoch": 18.620055932880543,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 46606
    },
    {
      "epoch": 18.620455453455854,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 46607
    },
    {
      "epoch": 18.62085497403116,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 46608
    },
    {
      "epoch": 18.62125449460647,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.037,
      "step": 46609
    },
    {
      "epoch": 18.621654015181782,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 46610
    },
    {
      "epoch": 18.622053535757093,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46611
    },
    {
      "epoch": 18.6224530563324,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 46612
    },
    {
      "epoch": 18.62285257690771,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 46613
    },
    {
      "epoch": 18.62325209748302,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 46614
    },
    {
      "epoch": 18.62365161805833,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 46615
    },
    {
      "epoch": 18.62405113863364,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 46616
    },
    {
      "epoch": 18.62445065920895,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 46617
    },
    {
      "epoch": 18.62485017978426,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 46618
    },
    {
      "epoch": 18.625249700359568,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 46619
    },
    {
      "epoch": 18.62564922093488,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 46620
    },
    {
      "epoch": 18.62604874151019,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 46621
    },
    {
      "epoch": 18.626448262085496,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 46622
    },
    {
      "epoch": 18.626847782660807,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 46623
    },
    {
      "epoch": 18.627247303236118,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 46624
    },
    {
      "epoch": 18.627646823811425,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 46625
    },
    {
      "epoch": 18.628046344386735,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 46626
    },
    {
      "epoch": 18.628445864962046,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 46627
    },
    {
      "epoch": 18.628845385537357,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 46628
    },
    {
      "epoch": 18.629244906112664,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 46629
    },
    {
      "epoch": 18.629644426687975,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 46630
    },
    {
      "epoch": 18.630043947263285,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 46631
    },
    {
      "epoch": 18.630443467838592,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.036,
      "step": 46632
    },
    {
      "epoch": 18.630842988413903,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 46633
    },
    {
      "epoch": 18.631242508989214,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 46634
    },
    {
      "epoch": 18.63164202956452,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 46635
    },
    {
      "epoch": 18.63204155013983,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 46636
    },
    {
      "epoch": 18.632441070715142,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0097,
      "step": 46637
    },
    {
      "epoch": 18.632840591290453,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 46638
    },
    {
      "epoch": 18.63324011186576,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 46639
    },
    {
      "epoch": 18.63363963244107,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 46640
    },
    {
      "epoch": 18.63403915301638,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 46641
    },
    {
      "epoch": 18.63443867359169,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 46642
    },
    {
      "epoch": 18.634838194167,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 46643
    },
    {
      "epoch": 18.63523771474231,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 46644
    },
    {
      "epoch": 18.63563723531762,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 46645
    },
    {
      "epoch": 18.636036755892928,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 46646
    },
    {
      "epoch": 18.63643627646824,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 46647
    },
    {
      "epoch": 18.63683579704355,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0135,
      "step": 46648
    },
    {
      "epoch": 18.637235317618856,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 46649
    },
    {
      "epoch": 18.637634838194167,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 46650
    },
    {
      "epoch": 18.638034358769477,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 46651
    },
    {
      "epoch": 18.638433879344785,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 46652
    },
    {
      "epoch": 18.638833399920095,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 46653
    },
    {
      "epoch": 18.639232920495406,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 46654
    },
    {
      "epoch": 18.639632441070717,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 46655
    },
    {
      "epoch": 18.640031961646024,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 46656
    },
    {
      "epoch": 18.640431482221334,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 46657
    },
    {
      "epoch": 18.640831002796645,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 46658
    },
    {
      "epoch": 18.641230523371952,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 46659
    },
    {
      "epoch": 18.641630043947263,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 46660
    },
    {
      "epoch": 18.642029564522574,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 46661
    },
    {
      "epoch": 18.642429085097884,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 46662
    },
    {
      "epoch": 18.64282860567319,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 46663
    },
    {
      "epoch": 18.643228126248502,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 46664
    },
    {
      "epoch": 18.643627646823813,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 46665
    },
    {
      "epoch": 18.64402716739912,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 46666
    },
    {
      "epoch": 18.64442668797443,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 46667
    },
    {
      "epoch": 18.64482620854974,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 46668
    },
    {
      "epoch": 18.64522572912505,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 46669
    },
    {
      "epoch": 18.64562524970036,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 46670
    },
    {
      "epoch": 18.64602477027567,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 46671
    },
    {
      "epoch": 18.64642429085098,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 46672
    },
    {
      "epoch": 18.646823811426287,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 46673
    },
    {
      "epoch": 18.647223332001598,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 46674
    },
    {
      "epoch": 18.64762285257691,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 46675
    },
    {
      "epoch": 18.648022373152216,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 46676
    },
    {
      "epoch": 18.648421893727527,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 46677
    },
    {
      "epoch": 18.648821414302837,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 46678
    },
    {
      "epoch": 18.649220934878148,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 46679
    },
    {
      "epoch": 18.649620455453455,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 46680
    },
    {
      "epoch": 18.650019976028766,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 46681
    },
    {
      "epoch": 18.650419496604076,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 46682
    },
    {
      "epoch": 18.650819017179384,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 46683
    },
    {
      "epoch": 18.651218537754694,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 46684
    },
    {
      "epoch": 18.651618058330005,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 46685
    },
    {
      "epoch": 18.652017578905312,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 46686
    },
    {
      "epoch": 18.652417099480623,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 46687
    },
    {
      "epoch": 18.652816620055933,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 46688
    },
    {
      "epoch": 18.653216140631244,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 46689
    },
    {
      "epoch": 18.65361566120655,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 46690
    },
    {
      "epoch": 18.654015181781862,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 46691
    },
    {
      "epoch": 18.654414702357172,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0078,
      "step": 46692
    },
    {
      "epoch": 18.65481422293248,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0437,
      "step": 46693
    },
    {
      "epoch": 18.65521374350779,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 46694
    },
    {
      "epoch": 18.6556132640831,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 46695
    },
    {
      "epoch": 18.65601278465841,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 46696
    },
    {
      "epoch": 18.65641230523372,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 46697
    },
    {
      "epoch": 18.65681182580903,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 46698
    },
    {
      "epoch": 18.65721134638434,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 46699
    },
    {
      "epoch": 18.657610866959647,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 46700
    },
    {
      "epoch": 18.658010387534958,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 46701
    },
    {
      "epoch": 18.65840990811027,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 46702
    },
    {
      "epoch": 18.658809428685576,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 46703
    },
    {
      "epoch": 18.659208949260886,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 46704
    },
    {
      "epoch": 18.659608469836197,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 46705
    },
    {
      "epoch": 18.660007990411508,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 46706
    },
    {
      "epoch": 18.660407510986815,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 46707
    },
    {
      "epoch": 18.660807031562126,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 46708
    },
    {
      "epoch": 18.661206552137436,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 46709
    },
    {
      "epoch": 18.661606072712743,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 46710
    },
    {
      "epoch": 18.662005593288054,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 46711
    },
    {
      "epoch": 18.662405113863365,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 46712
    },
    {
      "epoch": 18.66280463443867,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 46713
    },
    {
      "epoch": 18.663204155013982,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 46714
    },
    {
      "epoch": 18.663603675589293,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 46715
    },
    {
      "epoch": 18.664003196164604,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 46716
    },
    {
      "epoch": 18.66440271673991,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 46717
    },
    {
      "epoch": 18.66480223731522,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 46718
    },
    {
      "epoch": 18.665201757890532,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 46719
    },
    {
      "epoch": 18.66560127846584,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 46720
    },
    {
      "epoch": 18.66600079904115,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 46721
    },
    {
      "epoch": 18.66640031961646,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 46722
    },
    {
      "epoch": 18.66679984019177,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 46723
    },
    {
      "epoch": 18.66719936076708,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 46724
    },
    {
      "epoch": 18.66759888134239,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 46725
    },
    {
      "epoch": 18.6679984019177,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 46726
    },
    {
      "epoch": 18.668397922493007,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 46727
    },
    {
      "epoch": 18.668797443068318,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 46728
    },
    {
      "epoch": 18.66919696364363,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 46729
    },
    {
      "epoch": 18.669596484218935,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 46730
    },
    {
      "epoch": 18.669996004794246,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 46731
    },
    {
      "epoch": 18.670395525369557,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0363,
      "step": 46732
    },
    {
      "epoch": 18.670795045944867,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0115,
      "step": 46733
    },
    {
      "epoch": 18.671194566520175,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 46734
    },
    {
      "epoch": 18.671594087095485,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 46735
    },
    {
      "epoch": 18.671993607670796,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0107,
      "step": 46736
    },
    {
      "epoch": 18.672393128246103,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0082,
      "step": 46737
    },
    {
      "epoch": 18.672792648821414,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 46738
    },
    {
      "epoch": 18.673192169396724,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 46739
    },
    {
      "epoch": 18.673591689972035,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0112,
      "step": 46740
    },
    {
      "epoch": 18.673991210547342,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 46741
    },
    {
      "epoch": 18.674390731122653,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 46742
    },
    {
      "epoch": 18.674790251697964,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 46743
    },
    {
      "epoch": 18.67518977227327,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 46744
    },
    {
      "epoch": 18.67558929284858,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 46745
    },
    {
      "epoch": 18.675988813423892,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 46746
    },
    {
      "epoch": 18.6763883339992,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 46747
    },
    {
      "epoch": 18.67678785457451,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 46748
    },
    {
      "epoch": 18.67718737514982,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 46749
    },
    {
      "epoch": 18.67758689572513,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0005,
      "loss": 1.0121,
      "step": 46750
    },
    {
      "epoch": 18.67798641630044,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 46751
    },
    {
      "epoch": 18.67838593687575,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 46752
    },
    {
      "epoch": 18.67878545745106,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 46753
    },
    {
      "epoch": 18.679184978026367,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 46754
    },
    {
      "epoch": 18.679584498601677,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 46755
    },
    {
      "epoch": 18.679984019176988,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 46756
    },
    {
      "epoch": 18.6803835397523,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 46757
    },
    {
      "epoch": 18.680783060327606,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 46758
    },
    {
      "epoch": 18.681182580902917,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 46759
    },
    {
      "epoch": 18.681582101478227,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 46760
    },
    {
      "epoch": 18.681981622053534,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 46761
    },
    {
      "epoch": 18.682381142628845,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.009,
      "step": 46762
    },
    {
      "epoch": 18.682780663204156,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 46763
    },
    {
      "epoch": 18.683180183779463,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0073,
      "step": 46764
    },
    {
      "epoch": 18.683579704354774,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 46765
    },
    {
      "epoch": 18.683979224930084,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 46766
    },
    {
      "epoch": 18.684378745505395,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 46767
    },
    {
      "epoch": 18.684778266080702,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 46768
    },
    {
      "epoch": 18.685177786656013,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 46769
    },
    {
      "epoch": 18.685577307231323,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 46770
    },
    {
      "epoch": 18.68597682780663,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 46771
    },
    {
      "epoch": 18.68637634838194,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 46772
    },
    {
      "epoch": 18.686775868957252,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 46773
    },
    {
      "epoch": 18.687175389532563,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 46774
    },
    {
      "epoch": 18.68757491010787,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 46775
    },
    {
      "epoch": 18.68797443068318,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 46776
    },
    {
      "epoch": 18.68837395125849,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 46777
    },
    {
      "epoch": 18.688773471833798,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 46778
    },
    {
      "epoch": 18.68917299240911,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 46779
    },
    {
      "epoch": 18.68957251298442,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 46780
    },
    {
      "epoch": 18.689972033559727,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0118,
      "step": 46781
    },
    {
      "epoch": 18.690371554135037,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 46782
    },
    {
      "epoch": 18.690771074710348,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 46783
    },
    {
      "epoch": 18.69117059528566,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 46784
    },
    {
      "epoch": 18.691570115860966,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0098,
      "step": 46785
    },
    {
      "epoch": 18.691969636436276,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 46786
    },
    {
      "epoch": 18.692369157011587,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 46787
    },
    {
      "epoch": 18.692768677586894,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 46788
    },
    {
      "epoch": 18.693168198162205,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 46789
    },
    {
      "epoch": 18.693567718737516,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 46790
    },
    {
      "epoch": 18.693967239312826,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 46791
    },
    {
      "epoch": 18.694366759888133,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 46792
    },
    {
      "epoch": 18.694766280463444,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 46793
    },
    {
      "epoch": 18.695165801038755,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 46794
    },
    {
      "epoch": 18.695565321614062,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 46795
    },
    {
      "epoch": 18.695964842189372,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 46796
    },
    {
      "epoch": 18.696364362764683,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 46797
    },
    {
      "epoch": 18.69676388333999,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 46798
    },
    {
      "epoch": 18.6971634039153,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 46799
    },
    {
      "epoch": 18.69756292449061,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 46800
    },
    {
      "epoch": 18.697962445065922,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 46801
    },
    {
      "epoch": 18.69836196564123,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 46802
    },
    {
      "epoch": 18.69876148621654,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 46803
    },
    {
      "epoch": 18.69916100679185,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 46804
    },
    {
      "epoch": 18.699560527367158,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0091,
      "step": 46805
    },
    {
      "epoch": 18.69996004794247,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0104,
      "step": 46806
    },
    {
      "epoch": 18.70035956851778,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 46807
    },
    {
      "epoch": 18.70075908909309,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 46808
    },
    {
      "epoch": 18.701158609668397,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 46809
    },
    {
      "epoch": 18.701558130243708,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 46810
    },
    {
      "epoch": 18.70195765081902,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 46811
    },
    {
      "epoch": 18.702357171394326,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 46812
    },
    {
      "epoch": 18.702756691969636,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 46813
    },
    {
      "epoch": 18.703156212544947,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0359,
      "step": 46814
    },
    {
      "epoch": 18.703555733120254,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 46815
    },
    {
      "epoch": 18.703955253695565,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 46816
    },
    {
      "epoch": 18.704354774270875,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 46817
    },
    {
      "epoch": 18.704754294846186,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 46818
    },
    {
      "epoch": 18.705153815421493,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 46819
    },
    {
      "epoch": 18.705553335996804,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 46820
    },
    {
      "epoch": 18.705952856572114,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 46821
    },
    {
      "epoch": 18.70635237714742,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 46822
    },
    {
      "epoch": 18.706751897722732,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 46823
    },
    {
      "epoch": 18.707151418298043,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 46824
    },
    {
      "epoch": 18.707550938873354,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 46825
    },
    {
      "epoch": 18.70795045944866,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 46826
    },
    {
      "epoch": 18.70834998002397,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 46827
    },
    {
      "epoch": 18.708749500599282,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 46828
    },
    {
      "epoch": 18.70914902117459,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 46829
    },
    {
      "epoch": 18.7095485417499,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 46830
    },
    {
      "epoch": 18.70994806232521,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 46831
    },
    {
      "epoch": 18.710347582900518,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 46832
    },
    {
      "epoch": 18.71074710347583,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 46833
    },
    {
      "epoch": 18.71114662405114,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 46834
    },
    {
      "epoch": 18.71154614462645,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 46835
    },
    {
      "epoch": 18.711945665201757,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 46836
    },
    {
      "epoch": 18.712345185777068,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0463,
      "step": 46837
    },
    {
      "epoch": 18.712744706352378,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 46838
    },
    {
      "epoch": 18.713144226927685,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 46839
    },
    {
      "epoch": 18.713543747502996,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 46840
    },
    {
      "epoch": 18.713943268078307,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 46841
    },
    {
      "epoch": 18.714342788653617,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 46842
    },
    {
      "epoch": 18.714742309228924,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 46843
    },
    {
      "epoch": 18.715141829804235,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 46844
    },
    {
      "epoch": 18.715541350379546,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 46845
    },
    {
      "epoch": 18.715940870954853,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 46846
    },
    {
      "epoch": 18.716340391530164,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 46847
    },
    {
      "epoch": 18.716739912105474,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 46848
    },
    {
      "epoch": 18.71713943268078,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 46849
    },
    {
      "epoch": 18.717538953256092,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 46850
    },
    {
      "epoch": 18.717938473831403,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 46851
    },
    {
      "epoch": 18.718337994406713,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 46852
    },
    {
      "epoch": 18.71873751498202,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 46853
    },
    {
      "epoch": 18.71913703555733,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.012,
      "step": 46854
    },
    {
      "epoch": 18.719536556132642,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 46855
    },
    {
      "epoch": 18.71993607670795,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 46856
    },
    {
      "epoch": 18.72033559728326,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 46857
    },
    {
      "epoch": 18.72073511785857,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 46858
    },
    {
      "epoch": 18.72113463843388,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0427,
      "step": 46859
    },
    {
      "epoch": 18.721534159009188,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 46860
    },
    {
      "epoch": 18.7219336795845,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 46861
    },
    {
      "epoch": 18.72233320015981,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 46862
    },
    {
      "epoch": 18.722732720735117,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 46863
    },
    {
      "epoch": 18.723132241310427,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 46864
    },
    {
      "epoch": 18.723531761885738,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 46865
    },
    {
      "epoch": 18.723931282461045,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 46866
    },
    {
      "epoch": 18.724330803036356,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 46867
    },
    {
      "epoch": 18.724730323611666,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 46868
    },
    {
      "epoch": 18.725129844186977,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 46869
    },
    {
      "epoch": 18.725529364762284,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 46870
    },
    {
      "epoch": 18.725928885337595,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 46871
    },
    {
      "epoch": 18.726328405912906,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 46872
    },
    {
      "epoch": 18.726727926488213,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 46873
    },
    {
      "epoch": 18.727127447063523,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 46874
    },
    {
      "epoch": 18.727526967638834,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 46875
    },
    {
      "epoch": 18.727926488214145,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 46876
    },
    {
      "epoch": 18.728326008789452,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 46877
    },
    {
      "epoch": 18.728725529364763,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 46878
    },
    {
      "epoch": 18.729125049940073,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 46879
    },
    {
      "epoch": 18.72952457051538,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 46880
    },
    {
      "epoch": 18.72992409109069,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 46881
    },
    {
      "epoch": 18.730323611666,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 46882
    },
    {
      "epoch": 18.73072313224131,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 46883
    },
    {
      "epoch": 18.73112265281662,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46884
    },
    {
      "epoch": 18.73152217339193,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 46885
    },
    {
      "epoch": 18.73192169396724,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 46886
    },
    {
      "epoch": 18.732321214542548,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 46887
    },
    {
      "epoch": 18.73272073511786,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 46888
    },
    {
      "epoch": 18.73312025569317,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 46889
    },
    {
      "epoch": 18.733519776268476,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.009,
      "step": 46890
    },
    {
      "epoch": 18.733919296843787,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0088,
      "step": 46891
    },
    {
      "epoch": 18.734318817419098,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 46892
    },
    {
      "epoch": 18.73471833799441,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 46893
    },
    {
      "epoch": 18.735117858569716,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 46894
    },
    {
      "epoch": 18.735517379145026,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 46895
    },
    {
      "epoch": 18.735916899720337,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 46896
    },
    {
      "epoch": 18.736316420295644,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 46897
    },
    {
      "epoch": 18.736715940870955,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 46898
    },
    {
      "epoch": 18.737115461446265,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 46899
    },
    {
      "epoch": 18.737514982021573,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0128,
      "step": 46900
    },
    {
      "epoch": 18.737914502596883,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 46901
    },
    {
      "epoch": 18.738314023172194,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 46902
    },
    {
      "epoch": 18.738713543747505,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46903
    },
    {
      "epoch": 18.73911306432281,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 46904
    },
    {
      "epoch": 18.739512584898122,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0134,
      "step": 46905
    },
    {
      "epoch": 18.739912105473433,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 46906
    },
    {
      "epoch": 18.74031162604874,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 46907
    },
    {
      "epoch": 18.74071114662405,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 46908
    },
    {
      "epoch": 18.74111066719936,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 46909
    },
    {
      "epoch": 18.741510187774672,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 46910
    },
    {
      "epoch": 18.74190970834998,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 46911
    },
    {
      "epoch": 18.74230922892529,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 46912
    },
    {
      "epoch": 18.7427087495006,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 46913
    },
    {
      "epoch": 18.743108270075908,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 46914
    },
    {
      "epoch": 18.74350779065122,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 46915
    },
    {
      "epoch": 18.74390731122653,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 46916
    },
    {
      "epoch": 18.744306831801836,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 46917
    },
    {
      "epoch": 18.744706352377147,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 46918
    },
    {
      "epoch": 18.745105872952458,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 46919
    },
    {
      "epoch": 18.74550539352777,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 46920
    },
    {
      "epoch": 18.745904914103075,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 46921
    },
    {
      "epoch": 18.746304434678386,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 46922
    },
    {
      "epoch": 18.746703955253697,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 46923
    },
    {
      "epoch": 18.747103475829004,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 46924
    },
    {
      "epoch": 18.747502996404315,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 46925
    },
    {
      "epoch": 18.747902516979625,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 46926
    },
    {
      "epoch": 18.748302037554936,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0344,
      "step": 46927
    },
    {
      "epoch": 18.748701558130243,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 46928
    },
    {
      "epoch": 18.749101078705554,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 46929
    },
    {
      "epoch": 18.749500599280864,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 46930
    },
    {
      "epoch": 18.74990011985617,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0113,
      "step": 46931
    },
    {
      "epoch": 18.750299640431482,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 46932
    },
    {
      "epoch": 18.750699161006793,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 46933
    },
    {
      "epoch": 18.7510986815821,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 46934
    },
    {
      "epoch": 18.75149820215741,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 46935
    },
    {
      "epoch": 18.75189772273272,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 46936
    },
    {
      "epoch": 18.752297243308032,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0189,
      "step": 46937
    },
    {
      "epoch": 18.75269676388334,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 46938
    },
    {
      "epoch": 18.75309628445865,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 46939
    },
    {
      "epoch": 18.75349580503396,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 46940
    },
    {
      "epoch": 18.753895325609268,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 46941
    },
    {
      "epoch": 18.754294846184578,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 46942
    },
    {
      "epoch": 18.75469436675989,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 46943
    },
    {
      "epoch": 18.7550938873352,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 46944
    },
    {
      "epoch": 18.755493407910507,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0386,
      "step": 46945
    },
    {
      "epoch": 18.755892928485817,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 46946
    },
    {
      "epoch": 18.756292449061128,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 46947
    },
    {
      "epoch": 18.756691969636435,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 46948
    },
    {
      "epoch": 18.757091490211746,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46949
    },
    {
      "epoch": 18.757491010787056,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 46950
    },
    {
      "epoch": 18.757890531362364,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 46951
    },
    {
      "epoch": 18.758290051937674,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 46952
    },
    {
      "epoch": 18.758689572512985,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 46953
    },
    {
      "epoch": 18.759089093088296,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 46954
    },
    {
      "epoch": 18.759488613663603,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 46955
    },
    {
      "epoch": 18.759888134238913,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 46956
    },
    {
      "epoch": 18.760287654814224,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 46957
    },
    {
      "epoch": 18.76068717538953,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 46958
    },
    {
      "epoch": 18.761086695964842,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 46959
    },
    {
      "epoch": 18.761486216540153,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 46960
    },
    {
      "epoch": 18.761885737115463,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 46961
    },
    {
      "epoch": 18.76228525769077,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 46962
    },
    {
      "epoch": 18.76268477826608,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 46963
    },
    {
      "epoch": 18.76308429884139,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 46964
    },
    {
      "epoch": 18.7634838194167,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 46965
    },
    {
      "epoch": 18.76388333999201,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 46966
    },
    {
      "epoch": 18.76428286056732,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0136,
      "step": 46967
    },
    {
      "epoch": 18.764682381142627,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 46968
    },
    {
      "epoch": 18.765081901717938,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 46969
    },
    {
      "epoch": 18.76548142229325,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 46970
    },
    {
      "epoch": 18.76588094286856,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 46971
    },
    {
      "epoch": 18.766280463443866,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 46972
    },
    {
      "epoch": 18.766679984019177,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 46973
    },
    {
      "epoch": 18.767079504594488,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 46974
    },
    {
      "epoch": 18.767479025169795,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 46975
    },
    {
      "epoch": 18.767878545745106,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 46976
    },
    {
      "epoch": 18.768278066320416,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 46977
    },
    {
      "epoch": 18.768677586895727,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 46978
    },
    {
      "epoch": 18.769077107471034,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 46979
    },
    {
      "epoch": 18.769476628046345,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 46980
    },
    {
      "epoch": 18.769876148621655,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 46981
    },
    {
      "epoch": 18.770275669196963,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 46982
    },
    {
      "epoch": 18.770675189772273,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 46983
    },
    {
      "epoch": 18.771074710347584,
      "grad_norm": 0.037109375,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 46984
    },
    {
      "epoch": 18.77147423092289,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 46985
    },
    {
      "epoch": 18.7718737514982,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 46986
    },
    {
      "epoch": 18.772273272073512,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 46987
    },
    {
      "epoch": 18.772672792648823,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 46988
    },
    {
      "epoch": 18.77307231322413,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 46989
    },
    {
      "epoch": 18.77347183379944,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 46990
    },
    {
      "epoch": 18.77387135437475,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0276,
      "step": 46991
    },
    {
      "epoch": 18.77427087495006,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 46992
    },
    {
      "epoch": 18.77467039552537,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 46993
    },
    {
      "epoch": 18.77506991610068,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 46994
    },
    {
      "epoch": 18.775469436675987,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 46995
    },
    {
      "epoch": 18.775868957251298,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 46996
    },
    {
      "epoch": 18.77626847782661,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 46997
    },
    {
      "epoch": 18.77666799840192,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 46998
    },
    {
      "epoch": 18.777067518977226,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 46999
    },
    {
      "epoch": 18.777467039552537,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 47000
    },
    {
      "epoch": 18.777866560127848,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 47001
    },
    {
      "epoch": 18.778266080703155,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 47002
    },
    {
      "epoch": 18.778665601278465,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 47003
    },
    {
      "epoch": 18.779065121853776,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 47004
    },
    {
      "epoch": 18.779464642429087,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 47005
    },
    {
      "epoch": 18.779864163004394,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 47006
    },
    {
      "epoch": 18.780263683579705,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 47007
    },
    {
      "epoch": 18.780663204155015,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 47008
    },
    {
      "epoch": 18.781062724730322,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0371,
      "step": 47009
    },
    {
      "epoch": 18.781462245305633,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 47010
    },
    {
      "epoch": 18.781861765880944,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 47011
    },
    {
      "epoch": 18.78226128645625,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 47012
    },
    {
      "epoch": 18.78266080703156,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0131,
      "step": 47013
    },
    {
      "epoch": 18.783060327606872,
      "grad_norm": 0.125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 47014
    },
    {
      "epoch": 18.783459848182183,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 47015
    },
    {
      "epoch": 18.78385936875749,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 47016
    },
    {
      "epoch": 18.7842588893328,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 47017
    },
    {
      "epoch": 18.78465840990811,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 47018
    },
    {
      "epoch": 18.78505793048342,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 47019
    },
    {
      "epoch": 18.78545745105873,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0329,
      "step": 47020
    },
    {
      "epoch": 18.78585697163404,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0115,
      "step": 47021
    },
    {
      "epoch": 18.78625649220935,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 47022
    },
    {
      "epoch": 18.786656012784658,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 47023
    },
    {
      "epoch": 18.78705553335997,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 47024
    },
    {
      "epoch": 18.78745505393528,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 47025
    },
    {
      "epoch": 18.787854574510586,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 47026
    },
    {
      "epoch": 18.788254095085897,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 47027
    },
    {
      "epoch": 18.788653615661207,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 47028
    },
    {
      "epoch": 18.789053136236515,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 47029
    },
    {
      "epoch": 18.789452656811825,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 47030
    },
    {
      "epoch": 18.789852177387136,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 47031
    },
    {
      "epoch": 18.790251697962447,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 47032
    },
    {
      "epoch": 18.790651218537754,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 47033
    },
    {
      "epoch": 18.791050739113064,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 47034
    },
    {
      "epoch": 18.791450259688375,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 47035
    },
    {
      "epoch": 18.791849780263682,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 47036
    },
    {
      "epoch": 18.792249300838993,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 47037
    },
    {
      "epoch": 18.792648821414303,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 47038
    },
    {
      "epoch": 18.793048341989614,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 47039
    },
    {
      "epoch": 18.79344786256492,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0372,
      "step": 47040
    },
    {
      "epoch": 18.793847383140232,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 47041
    },
    {
      "epoch": 18.794246903715543,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 47042
    },
    {
      "epoch": 18.79464642429085,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 47043
    },
    {
      "epoch": 18.79504594486616,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 47044
    },
    {
      "epoch": 18.79544546544147,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 47045
    },
    {
      "epoch": 18.795844986016778,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0005,
      "loss": 1.0059,
      "step": 47046
    },
    {
      "epoch": 18.79624450659209,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 47047
    },
    {
      "epoch": 18.7966440271674,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 47048
    },
    {
      "epoch": 18.79704354774271,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 47049
    },
    {
      "epoch": 18.797443068318017,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 47050
    },
    {
      "epoch": 18.797842588893328,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0119,
      "step": 47051
    },
    {
      "epoch": 18.79824210946864,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 47052
    },
    {
      "epoch": 18.798641630043946,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 47053
    },
    {
      "epoch": 18.799041150619257,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 47054
    },
    {
      "epoch": 18.799440671194567,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 47055
    },
    {
      "epoch": 18.799840191769878,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 47056
    },
    {
      "epoch": 18.800239712345185,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 47057
    },
    {
      "epoch": 18.800639232920496,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 47058
    },
    {
      "epoch": 18.801038753495806,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0357,
      "step": 47059
    },
    {
      "epoch": 18.801438274071113,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 47060
    },
    {
      "epoch": 18.801837794646424,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 47061
    },
    {
      "epoch": 18.802237315221735,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 47062
    },
    {
      "epoch": 18.802636835797042,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 47063
    },
    {
      "epoch": 18.803036356372353,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0106,
      "step": 47064
    },
    {
      "epoch": 18.803435876947663,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 47065
    },
    {
      "epoch": 18.803835397522974,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 47066
    },
    {
      "epoch": 18.80423491809828,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 47067
    },
    {
      "epoch": 18.80463443867359,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 47068
    },
    {
      "epoch": 18.805033959248902,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 47069
    },
    {
      "epoch": 18.80543347982421,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 47070
    },
    {
      "epoch": 18.80583300039952,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 47071
    },
    {
      "epoch": 18.80623252097483,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 47072
    },
    {
      "epoch": 18.806632041550138,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 47073
    },
    {
      "epoch": 18.80703156212545,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 47074
    },
    {
      "epoch": 18.80743108270076,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 47075
    },
    {
      "epoch": 18.80783060327607,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 47076
    },
    {
      "epoch": 18.808230123851377,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 47077
    },
    {
      "epoch": 18.808629644426688,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0128,
      "step": 47078
    },
    {
      "epoch": 18.809029165002,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 47079
    },
    {
      "epoch": 18.809428685577306,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 47080
    },
    {
      "epoch": 18.809828206152616,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0418,
      "step": 47081
    },
    {
      "epoch": 18.810227726727927,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 47082
    },
    {
      "epoch": 18.810627247303238,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 47083
    },
    {
      "epoch": 18.811026767878545,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 47084
    },
    {
      "epoch": 18.811426288453855,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 47085
    },
    {
      "epoch": 18.811825809029166,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 47086
    },
    {
      "epoch": 18.812225329604473,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 47087
    },
    {
      "epoch": 18.812624850179784,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0148,
      "step": 47088
    },
    {
      "epoch": 18.813024370755095,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 47089
    },
    {
      "epoch": 18.8134238913304,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 47090
    },
    {
      "epoch": 18.813823411905712,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 47091
    },
    {
      "epoch": 18.814222932481023,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 47092
    },
    {
      "epoch": 18.814622453056334,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 47093
    },
    {
      "epoch": 18.81502197363164,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 47094
    },
    {
      "epoch": 18.81542149420695,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 47095
    },
    {
      "epoch": 18.815821014782262,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 47096
    },
    {
      "epoch": 18.81622053535757,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 47097
    },
    {
      "epoch": 18.81662005593288,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 47098
    },
    {
      "epoch": 18.81701957650819,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 47099
    },
    {
      "epoch": 18.8174190970835,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 47100
    },
    {
      "epoch": 18.81781861765881,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 47101
    },
    {
      "epoch": 18.81821813823412,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 47102
    },
    {
      "epoch": 18.81861765880943,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 47103
    },
    {
      "epoch": 18.819017179384737,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 47104
    },
    {
      "epoch": 18.819416699960048,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 47105
    },
    {
      "epoch": 18.81981622053536,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 47106
    },
    {
      "epoch": 18.820215741110665,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 47107
    },
    {
      "epoch": 18.820615261685976,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 47108
    },
    {
      "epoch": 18.821014782261287,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 47109
    },
    {
      "epoch": 18.821414302836597,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 47110
    },
    {
      "epoch": 18.821813823411905,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 47111
    },
    {
      "epoch": 18.822213343987215,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 47112
    },
    {
      "epoch": 18.822612864562526,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 47113
    },
    {
      "epoch": 18.823012385137833,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 47114
    },
    {
      "epoch": 18.823411905713144,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 47115
    },
    {
      "epoch": 18.823811426288454,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0089,
      "step": 47116
    },
    {
      "epoch": 18.824210946863765,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 47117
    },
    {
      "epoch": 18.824610467439072,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 47118
    },
    {
      "epoch": 18.825009988014383,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 47119
    },
    {
      "epoch": 18.825409508589694,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 47120
    },
    {
      "epoch": 18.825809029165,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 47121
    },
    {
      "epoch": 18.82620854974031,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 47122
    },
    {
      "epoch": 18.826608070315622,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 47123
    },
    {
      "epoch": 18.82700759089093,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 47124
    },
    {
      "epoch": 18.82740711146624,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 47125
    },
    {
      "epoch": 18.82780663204155,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 47126
    },
    {
      "epoch": 18.82820615261686,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 47127
    },
    {
      "epoch": 18.82860567319217,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.041,
      "step": 47128
    },
    {
      "epoch": 18.82900519376748,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 47129
    },
    {
      "epoch": 18.82940471434279,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 47130
    },
    {
      "epoch": 18.829804234918097,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 47131
    },
    {
      "epoch": 18.830203755493407,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 47132
    },
    {
      "epoch": 18.830603276068718,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 47133
    },
    {
      "epoch": 18.83100279664403,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 47134
    },
    {
      "epoch": 18.831402317219336,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 47135
    },
    {
      "epoch": 18.831801837794647,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 47136
    },
    {
      "epoch": 18.832201358369957,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 47137
    },
    {
      "epoch": 18.832600878945264,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 47138
    },
    {
      "epoch": 18.833000399520575,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 47139
    },
    {
      "epoch": 18.833399920095886,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0397,
      "step": 47140
    },
    {
      "epoch": 18.833799440671193,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 47141
    },
    {
      "epoch": 18.834198961246504,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 47142
    },
    {
      "epoch": 18.834598481821814,
      "grad_norm": 0.060546875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 47143
    },
    {
      "epoch": 18.834998002397125,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 47144
    },
    {
      "epoch": 18.835397522972432,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 47145
    },
    {
      "epoch": 18.835797043547743,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 47146
    },
    {
      "epoch": 18.836196564123053,
      "grad_norm": 0.03662109375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 47147
    },
    {
      "epoch": 18.83659608469836,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0365,
      "step": 47148
    },
    {
      "epoch": 18.83699560527367,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 47149
    },
    {
      "epoch": 18.837395125848982,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 47150
    },
    {
      "epoch": 18.837794646424292,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 47151
    },
    {
      "epoch": 18.8381941669996,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 47152
    },
    {
      "epoch": 18.83859368757491,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 47153
    },
    {
      "epoch": 18.83899320815022,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 47154
    },
    {
      "epoch": 18.839392728725528,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 47155
    },
    {
      "epoch": 18.83979224930084,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 47156
    },
    {
      "epoch": 18.84019176987615,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 47157
    },
    {
      "epoch": 18.840591290451457,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 47158
    },
    {
      "epoch": 18.840990811026767,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.009,
      "step": 47159
    },
    {
      "epoch": 18.841390331602078,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 47160
    },
    {
      "epoch": 18.84178985217739,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 47161
    },
    {
      "epoch": 18.842189372752696,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 47162
    },
    {
      "epoch": 18.842588893328006,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 47163
    },
    {
      "epoch": 18.842988413903317,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 47164
    },
    {
      "epoch": 18.843387934478624,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 47165
    },
    {
      "epoch": 18.843787455053935,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 47166
    },
    {
      "epoch": 18.844186975629245,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 47167
    },
    {
      "epoch": 18.844586496204556,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 47168
    },
    {
      "epoch": 18.844986016779863,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 47169
    },
    {
      "epoch": 18.845385537355174,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 47170
    },
    {
      "epoch": 18.845785057930485,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 47171
    },
    {
      "epoch": 18.84618457850579,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 47172
    },
    {
      "epoch": 18.846584099081102,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 47173
    },
    {
      "epoch": 18.846983619656413,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 47174
    },
    {
      "epoch": 18.84738314023172,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 47175
    },
    {
      "epoch": 18.84778266080703,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 47176
    },
    {
      "epoch": 18.84818218138234,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 47177
    },
    {
      "epoch": 18.848581701957652,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 47178
    },
    {
      "epoch": 18.84898122253296,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 47179
    },
    {
      "epoch": 18.84938074310827,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0367,
      "step": 47180
    },
    {
      "epoch": 18.84978026368358,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 47181
    },
    {
      "epoch": 18.850179784258888,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 47182
    },
    {
      "epoch": 18.8505793048342,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 47183
    },
    {
      "epoch": 18.85097882540951,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 47184
    },
    {
      "epoch": 18.85137834598482,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 47185
    },
    {
      "epoch": 18.851777866560127,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 47186
    },
    {
      "epoch": 18.852177387135438,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 47187
    },
    {
      "epoch": 18.85257690771075,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0062,
      "step": 47188
    },
    {
      "epoch": 18.852976428286055,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 47189
    },
    {
      "epoch": 18.853375948861366,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 47190
    },
    {
      "epoch": 18.853775469436677,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0154,
      "step": 47191
    },
    {
      "epoch": 18.854174990011984,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 47192
    },
    {
      "epoch": 18.854574510587295,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 47193
    },
    {
      "epoch": 18.854974031162605,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 47194
    },
    {
      "epoch": 18.855373551737916,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 47195
    },
    {
      "epoch": 18.855773072313223,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 47196
    },
    {
      "epoch": 18.856172592888534,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 47197
    },
    {
      "epoch": 18.856572113463844,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 47198
    },
    {
      "epoch": 18.85697163403915,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 47199
    },
    {
      "epoch": 18.857371154614462,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0187,
      "step": 47200
    },
    {
      "epoch": 18.857770675189773,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 47201
    },
    {
      "epoch": 18.858170195765084,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 47202
    },
    {
      "epoch": 18.85856971634039,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0104,
      "step": 47203
    },
    {
      "epoch": 18.8589692369157,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 47204
    },
    {
      "epoch": 18.859368757491012,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 47205
    },
    {
      "epoch": 18.85976827806632,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 47206
    },
    {
      "epoch": 18.86016779864163,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 47207
    },
    {
      "epoch": 18.86056731921694,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 47208
    },
    {
      "epoch": 18.860966839792248,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 47209
    },
    {
      "epoch": 18.86136636036756,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 47210
    },
    {
      "epoch": 18.86176588094287,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 47211
    },
    {
      "epoch": 18.86216540151818,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 47212
    },
    {
      "epoch": 18.862564922093487,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 47213
    },
    {
      "epoch": 18.862964442668797,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0168,
      "step": 47214
    },
    {
      "epoch": 18.863363963244108,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 47215
    },
    {
      "epoch": 18.863763483819415,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0389,
      "step": 47216
    },
    {
      "epoch": 18.864163004394726,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 47217
    },
    {
      "epoch": 18.864562524970037,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 47218
    },
    {
      "epoch": 18.864962045545347,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0125,
      "step": 47219
    },
    {
      "epoch": 18.865361566120654,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 47220
    },
    {
      "epoch": 18.865761086695965,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 47221
    },
    {
      "epoch": 18.866160607271276,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 47222
    },
    {
      "epoch": 18.866560127846583,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 47223
    },
    {
      "epoch": 18.866959648421894,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 47224
    },
    {
      "epoch": 18.867359168997204,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 47225
    },
    {
      "epoch": 18.86775868957251,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 47226
    },
    {
      "epoch": 18.868158210147822,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 47227
    },
    {
      "epoch": 18.868557730723133,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 47228
    },
    {
      "epoch": 18.868957251298443,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0081,
      "step": 47229
    },
    {
      "epoch": 18.86935677187375,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 47230
    },
    {
      "epoch": 18.86975629244906,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 47231
    },
    {
      "epoch": 18.870155813024372,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 47232
    },
    {
      "epoch": 18.87055533359968,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 47233
    },
    {
      "epoch": 18.87095485417499,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 47234
    },
    {
      "epoch": 18.8713543747503,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 47235
    },
    {
      "epoch": 18.87175389532561,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 47236
    },
    {
      "epoch": 18.872153415900918,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 47237
    },
    {
      "epoch": 18.87255293647623,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 47238
    },
    {
      "epoch": 18.87295245705154,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 47239
    },
    {
      "epoch": 18.873351977626847,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 47240
    },
    {
      "epoch": 18.873751498202157,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 47241
    },
    {
      "epoch": 18.874151018777468,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 47242
    },
    {
      "epoch": 18.874550539352775,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 47243
    },
    {
      "epoch": 18.874950059928086,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 47244
    },
    {
      "epoch": 18.875349580503396,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 47245
    },
    {
      "epoch": 18.875749101078707,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 47246
    },
    {
      "epoch": 18.876148621654014,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 47247
    },
    {
      "epoch": 18.876548142229325,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 47248
    },
    {
      "epoch": 18.876947662804636,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 47249
    },
    {
      "epoch": 18.877347183379943,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0125,
      "step": 47250
    },
    {
      "epoch": 18.877746703955253,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 47251
    },
    {
      "epoch": 18.878146224530564,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.015,
      "step": 47252
    },
    {
      "epoch": 18.878545745105875,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 47253
    },
    {
      "epoch": 18.878945265681182,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 47254
    },
    {
      "epoch": 18.879344786256492,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 47255
    },
    {
      "epoch": 18.879744306831803,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 47256
    },
    {
      "epoch": 18.88014382740711,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0343,
      "step": 47257
    },
    {
      "epoch": 18.88054334798242,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 47258
    },
    {
      "epoch": 18.88094286855773,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 47259
    },
    {
      "epoch": 18.88134238913304,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 47260
    },
    {
      "epoch": 18.88174190970835,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 47261
    },
    {
      "epoch": 18.88214143028366,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 47262
    },
    {
      "epoch": 18.88254095085897,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 47263
    },
    {
      "epoch": 18.882940471434278,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 47264
    },
    {
      "epoch": 18.88333999200959,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 47265
    },
    {
      "epoch": 18.8837395125849,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0379,
      "step": 47266
    },
    {
      "epoch": 18.884139033160206,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 47267
    },
    {
      "epoch": 18.884538553735517,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 47268
    },
    {
      "epoch": 18.884938074310828,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 47269
    },
    {
      "epoch": 18.88533759488614,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 47270
    },
    {
      "epoch": 18.885737115461446,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 47271
    },
    {
      "epoch": 18.886136636036756,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 47272
    },
    {
      "epoch": 18.886536156612067,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0005,
      "loss": 1.0122,
      "step": 47273
    },
    {
      "epoch": 18.886935677187374,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 47274
    },
    {
      "epoch": 18.887335197762685,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 47275
    },
    {
      "epoch": 18.887734718337995,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 47276
    },
    {
      "epoch": 18.888134238913302,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 47277
    },
    {
      "epoch": 18.888533759488613,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 47278
    },
    {
      "epoch": 18.888933280063924,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 47279
    },
    {
      "epoch": 18.889332800639234,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 47280
    },
    {
      "epoch": 18.88973232121454,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 47281
    },
    {
      "epoch": 18.890131841789852,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 47282
    },
    {
      "epoch": 18.890531362365163,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0119,
      "step": 47283
    },
    {
      "epoch": 18.89093088294047,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.03,
      "step": 47284
    },
    {
      "epoch": 18.89133040351578,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0139,
      "step": 47285
    },
    {
      "epoch": 18.89172992409109,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 47286
    },
    {
      "epoch": 18.892129444666402,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 47287
    },
    {
      "epoch": 18.89252896524171,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 47288
    },
    {
      "epoch": 18.89292848581702,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 47289
    },
    {
      "epoch": 18.89332800639233,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 47290
    },
    {
      "epoch": 18.893727526967638,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0325,
      "step": 47291
    },
    {
      "epoch": 18.89412704754295,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 47292
    },
    {
      "epoch": 18.89452656811826,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 47293
    },
    {
      "epoch": 18.894926088693566,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 47294
    },
    {
      "epoch": 18.895325609268877,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0267,
      "step": 47295
    },
    {
      "epoch": 18.895725129844188,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 47296
    },
    {
      "epoch": 18.896124650419498,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 47297
    },
    {
      "epoch": 18.896524170994805,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0337,
      "step": 47298
    },
    {
      "epoch": 18.896923691570116,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 47299
    },
    {
      "epoch": 18.897323212145427,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 47300
    },
    {
      "epoch": 18.897722732720734,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 47301
    },
    {
      "epoch": 18.898122253296044,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 47302
    },
    {
      "epoch": 18.898521773871355,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 47303
    },
    {
      "epoch": 18.898921294446666,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 47304
    },
    {
      "epoch": 18.899320815021973,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 47305
    },
    {
      "epoch": 18.899720335597284,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 47306
    },
    {
      "epoch": 18.900119856172594,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 47307
    },
    {
      "epoch": 18.9005193767479,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 47308
    },
    {
      "epoch": 18.900918897323212,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 47309
    },
    {
      "epoch": 18.901318417898523,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 47310
    },
    {
      "epoch": 18.90171793847383,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 47311
    },
    {
      "epoch": 18.90211745904914,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0328,
      "step": 47312
    },
    {
      "epoch": 18.90251697962445,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0114,
      "step": 47313
    },
    {
      "epoch": 18.902916500199762,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 47314
    },
    {
      "epoch": 18.90331602077507,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 47315
    },
    {
      "epoch": 18.90371554135038,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 47316
    },
    {
      "epoch": 18.90411506192569,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 47317
    },
    {
      "epoch": 18.904514582500997,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 47318
    },
    {
      "epoch": 18.904914103076308,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0072,
      "step": 47319
    },
    {
      "epoch": 18.90531362365162,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 47320
    },
    {
      "epoch": 18.90571314422693,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 47321
    },
    {
      "epoch": 18.906112664802237,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 47322
    },
    {
      "epoch": 18.906512185377547,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 47323
    },
    {
      "epoch": 18.906911705952858,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 47324
    },
    {
      "epoch": 18.907311226528165,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 47325
    },
    {
      "epoch": 18.907710747103476,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 47326
    },
    {
      "epoch": 18.908110267678786,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 47327
    },
    {
      "epoch": 18.908509788254094,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 47328
    },
    {
      "epoch": 18.908909308829404,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 47329
    },
    {
      "epoch": 18.909308829404715,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 47330
    },
    {
      "epoch": 18.909708349980026,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0409,
      "step": 47331
    },
    {
      "epoch": 18.910107870555333,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 47332
    },
    {
      "epoch": 18.910507391130643,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 47333
    },
    {
      "epoch": 18.910906911705954,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0005,
      "loss": 1.0093,
      "step": 47334
    },
    {
      "epoch": 18.91130643228126,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 47335
    },
    {
      "epoch": 18.911705952856572,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 47336
    },
    {
      "epoch": 18.912105473431883,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 47337
    },
    {
      "epoch": 18.91250499400719,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 47338
    },
    {
      "epoch": 18.9129045145825,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 47339
    },
    {
      "epoch": 18.91330403515781,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 47340
    },
    {
      "epoch": 18.91370355573312,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 47341
    },
    {
      "epoch": 18.91410307630843,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 47342
    },
    {
      "epoch": 18.91450259688374,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 47343
    },
    {
      "epoch": 18.91490211745905,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 47344
    },
    {
      "epoch": 18.915301638034357,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 47345
    },
    {
      "epoch": 18.915701158609668,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 47346
    },
    {
      "epoch": 18.91610067918498,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 47347
    },
    {
      "epoch": 18.91650019976029,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 47348
    },
    {
      "epoch": 18.916899720335596,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 47349
    },
    {
      "epoch": 18.917299240910907,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 47350
    },
    {
      "epoch": 18.917698761486218,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 47351
    },
    {
      "epoch": 18.918098282061525,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 47352
    },
    {
      "epoch": 18.918497802636836,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 47353
    },
    {
      "epoch": 18.918897323212146,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 47354
    },
    {
      "epoch": 18.919296843787453,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 47355
    },
    {
      "epoch": 18.919696364362764,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 47356
    },
    {
      "epoch": 18.920095884938075,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 47357
    },
    {
      "epoch": 18.920495405513385,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 47358
    },
    {
      "epoch": 18.920894926088693,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 47359
    },
    {
      "epoch": 18.921294446664003,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 47360
    },
    {
      "epoch": 18.921693967239314,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 47361
    },
    {
      "epoch": 18.92209348781462,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 47362
    },
    {
      "epoch": 18.92249300838993,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 47363
    },
    {
      "epoch": 18.922892528965242,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 47364
    },
    {
      "epoch": 18.923292049540553,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 47365
    },
    {
      "epoch": 18.92369157011586,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 47366
    },
    {
      "epoch": 18.92409109069117,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 47367
    },
    {
      "epoch": 18.92449061126648,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 47368
    },
    {
      "epoch": 18.92489013184179,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 47369
    },
    {
      "epoch": 18.9252896524171,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 47370
    },
    {
      "epoch": 18.92568917299241,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 47371
    },
    {
      "epoch": 18.926088693567717,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 47372
    },
    {
      "epoch": 18.926488214143028,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0257,
      "step": 47373
    },
    {
      "epoch": 18.92688773471834,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 47374
    },
    {
      "epoch": 18.92728725529365,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 47375
    },
    {
      "epoch": 18.927686775868956,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 47376
    },
    {
      "epoch": 18.928086296444267,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 47377
    },
    {
      "epoch": 18.928485817019578,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 47378
    },
    {
      "epoch": 18.928885337594885,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 47379
    },
    {
      "epoch": 18.929284858170195,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 47380
    },
    {
      "epoch": 18.929684378745506,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 47381
    },
    {
      "epoch": 18.930083899320817,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 47382
    },
    {
      "epoch": 18.930483419896124,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 47383
    },
    {
      "epoch": 18.930882940471434,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 47384
    },
    {
      "epoch": 18.931282461046745,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 47385
    },
    {
      "epoch": 18.931681981622052,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 47386
    },
    {
      "epoch": 18.932081502197363,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 47387
    },
    {
      "epoch": 18.932481022772674,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 47388
    },
    {
      "epoch": 18.93288054334798,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 47389
    },
    {
      "epoch": 18.93328006392329,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 47390
    },
    {
      "epoch": 18.933679584498602,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 47391
    },
    {
      "epoch": 18.934079105073913,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 47392
    },
    {
      "epoch": 18.93447862564922,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 47393
    },
    {
      "epoch": 18.93487814622453,
      "grad_norm": 0.171875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 47394
    },
    {
      "epoch": 18.93527766679984,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 47395
    },
    {
      "epoch": 18.93567718737515,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 47396
    },
    {
      "epoch": 18.93607670795046,
      "grad_norm": 0.15625,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 47397
    },
    {
      "epoch": 18.93647622852577,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 47398
    },
    {
      "epoch": 18.93687574910108,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 47399
    },
    {
      "epoch": 18.937275269676388,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 47400
    },
    {
      "epoch": 18.937674790251698,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 47401
    },
    {
      "epoch": 18.93807431082701,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 47402
    },
    {
      "epoch": 18.938473831402316,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 47403
    },
    {
      "epoch": 18.938873351977627,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 47404
    },
    {
      "epoch": 18.939272872552937,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 47405
    },
    {
      "epoch": 18.939672393128244,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 47406
    },
    {
      "epoch": 18.940071913703555,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 47407
    },
    {
      "epoch": 18.940471434278866,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 47408
    },
    {
      "epoch": 18.940870954854176,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0005,
      "loss": 1.0375,
      "step": 47409
    },
    {
      "epoch": 18.941270475429484,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 47410
    },
    {
      "epoch": 18.941669996004794,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 47411
    },
    {
      "epoch": 18.942069516580105,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 47412
    },
    {
      "epoch": 18.942469037155412,
      "grad_norm": 0.038330078125,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 47413
    },
    {
      "epoch": 18.942868557730723,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 47414
    },
    {
      "epoch": 18.943268078306033,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 47415
    },
    {
      "epoch": 18.943667598881344,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 47416
    },
    {
      "epoch": 18.94406711945665,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 47417
    },
    {
      "epoch": 18.944466640031962,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 47418
    },
    {
      "epoch": 18.944866160607273,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 47419
    },
    {
      "epoch": 18.94526568118258,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 47420
    },
    {
      "epoch": 18.94566520175789,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 47421
    },
    {
      "epoch": 18.9460647223332,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0218,
      "step": 47422
    },
    {
      "epoch": 18.946464242908508,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 47423
    },
    {
      "epoch": 18.94686376348382,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0294,
      "step": 47424
    },
    {
      "epoch": 18.94726328405913,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 47425
    },
    {
      "epoch": 18.94766280463444,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 47426
    },
    {
      "epoch": 18.948062325209747,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 47427
    },
    {
      "epoch": 18.948461845785058,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 47428
    },
    {
      "epoch": 18.94886136636037,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 47429
    },
    {
      "epoch": 18.949260886935676,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 47430
    },
    {
      "epoch": 18.949660407510986,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 47431
    },
    {
      "epoch": 18.950059928086297,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 47432
    },
    {
      "epoch": 18.950459448661604,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 47433
    },
    {
      "epoch": 18.950858969236915,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0161,
      "step": 47434
    },
    {
      "epoch": 18.951258489812226,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 47435
    },
    {
      "epoch": 18.951658010387536,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.0326,
      "step": 47436
    },
    {
      "epoch": 18.952057530962843,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 47437
    },
    {
      "epoch": 18.952457051538154,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 47438
    },
    {
      "epoch": 18.952856572113465,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0277,
      "step": 47439
    },
    {
      "epoch": 18.953256092688772,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 47440
    },
    {
      "epoch": 18.953655613264083,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0114,
      "step": 47441
    },
    {
      "epoch": 18.954055133839393,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 47442
    },
    {
      "epoch": 18.954454654414704,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 47443
    },
    {
      "epoch": 18.95485417499001,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 47444
    },
    {
      "epoch": 18.95525369556532,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 47445
    },
    {
      "epoch": 18.955653216140632,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0282,
      "step": 47446
    },
    {
      "epoch": 18.95605273671594,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 47447
    },
    {
      "epoch": 18.95645225729125,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 47448
    },
    {
      "epoch": 18.95685177786656,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 47449
    },
    {
      "epoch": 18.957251298441868,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 47450
    },
    {
      "epoch": 18.95765081901718,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 47451
    },
    {
      "epoch": 18.95805033959249,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 47452
    },
    {
      "epoch": 18.9584498601678,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 47453
    },
    {
      "epoch": 18.958849380743107,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 47454
    },
    {
      "epoch": 18.959248901318418,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 47455
    },
    {
      "epoch": 18.95964842189373,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 47456
    },
    {
      "epoch": 18.960047942469036,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 47457
    },
    {
      "epoch": 18.960447463044346,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 47458
    },
    {
      "epoch": 18.960846983619657,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 47459
    },
    {
      "epoch": 18.961246504194968,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 47460
    },
    {
      "epoch": 18.961646024770275,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 47461
    },
    {
      "epoch": 18.962045545345585,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 47462
    },
    {
      "epoch": 18.962445065920896,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 47463
    },
    {
      "epoch": 18.962844586496203,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 47464
    },
    {
      "epoch": 18.963244107071514,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0143,
      "step": 47465
    },
    {
      "epoch": 18.963643627646825,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 47466
    },
    {
      "epoch": 18.96404314822213,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 47467
    },
    {
      "epoch": 18.964442668797442,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 47468
    },
    {
      "epoch": 18.964842189372753,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 47469
    },
    {
      "epoch": 18.965241709948064,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0348,
      "step": 47470
    },
    {
      "epoch": 18.96564123052337,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0364,
      "step": 47471
    },
    {
      "epoch": 18.96604075109868,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0229,
      "step": 47472
    },
    {
      "epoch": 18.966440271673992,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0131,
      "step": 47473
    },
    {
      "epoch": 18.9668397922493,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 47474
    },
    {
      "epoch": 18.96723931282461,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 47475
    },
    {
      "epoch": 18.96763883339992,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 47476
    },
    {
      "epoch": 18.96803835397523,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 47477
    },
    {
      "epoch": 18.96843787455054,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 47478
    },
    {
      "epoch": 18.96883739512585,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 47479
    },
    {
      "epoch": 18.96923691570116,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 47480
    },
    {
      "epoch": 18.969636436276467,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 47481
    },
    {
      "epoch": 18.970035956851778,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 47482
    },
    {
      "epoch": 18.97043547742709,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0302,
      "step": 47483
    },
    {
      "epoch": 18.970834998002395,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 47484
    },
    {
      "epoch": 18.971234518577706,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 47485
    },
    {
      "epoch": 18.971634039153017,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 47486
    },
    {
      "epoch": 18.972033559728327,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 47487
    },
    {
      "epoch": 18.972433080303635,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 47488
    },
    {
      "epoch": 18.972832600878945,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 47489
    },
    {
      "epoch": 18.973232121454256,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 47490
    },
    {
      "epoch": 18.973631642029563,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 47491
    },
    {
      "epoch": 18.974031162604874,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 47492
    },
    {
      "epoch": 18.974430683180184,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 47493
    },
    {
      "epoch": 18.974830203755495,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0132,
      "step": 47494
    },
    {
      "epoch": 18.975229724330802,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 47495
    },
    {
      "epoch": 18.975629244906113,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 47496
    },
    {
      "epoch": 18.976028765481423,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 47497
    },
    {
      "epoch": 18.97642828605673,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 47498
    },
    {
      "epoch": 18.97682780663204,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 47499
    },
    {
      "epoch": 18.977227327207352,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 47500
    },
    {
      "epoch": 18.97762684778266,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 47501
    },
    {
      "epoch": 18.97802636835797,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0093,
      "step": 47502
    },
    {
      "epoch": 18.97842588893328,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0305,
      "step": 47503
    },
    {
      "epoch": 18.97882540950859,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 47504
    },
    {
      "epoch": 18.979224930083898,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.021,
      "step": 47505
    },
    {
      "epoch": 18.97962445065921,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 47506
    },
    {
      "epoch": 18.98002397123452,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0149,
      "step": 47507
    },
    {
      "epoch": 18.980423491809827,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 47508
    },
    {
      "epoch": 18.980823012385137,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 47509
    },
    {
      "epoch": 18.981222532960448,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0144,
      "step": 47510
    },
    {
      "epoch": 18.98162205353576,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 47511
    },
    {
      "epoch": 18.982021574111066,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 47512
    },
    {
      "epoch": 18.982421094686377,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0005,
      "loss": 1.0381,
      "step": 47513
    },
    {
      "epoch": 18.982820615261687,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 47514
    },
    {
      "epoch": 18.983220135836994,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 47515
    },
    {
      "epoch": 18.983619656412305,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 47516
    },
    {
      "epoch": 18.984019176987616,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 47517
    },
    {
      "epoch": 18.984418697562923,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 47518
    },
    {
      "epoch": 18.984818218138233,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 47519
    },
    {
      "epoch": 18.985217738713544,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 47520
    },
    {
      "epoch": 18.985617259288855,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0005,
      "loss": 1.0341,
      "step": 47521
    },
    {
      "epoch": 18.986016779864162,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0313,
      "step": 47522
    },
    {
      "epoch": 18.986416300439473,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 47523
    },
    {
      "epoch": 18.986815821014783,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 47524
    },
    {
      "epoch": 18.98721534159009,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 47525
    },
    {
      "epoch": 18.9876148621654,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 47526
    },
    {
      "epoch": 18.98801438274071,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 47527
    },
    {
      "epoch": 18.988413903316022,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0444,
      "step": 47528
    },
    {
      "epoch": 18.98881342389133,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0124,
      "step": 47529
    },
    {
      "epoch": 18.98921294446664,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 47530
    },
    {
      "epoch": 18.98961246504195,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 47531
    },
    {
      "epoch": 18.990011985617258,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 47532
    },
    {
      "epoch": 18.99041150619257,
      "grad_norm": 0.055419921875,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 47533
    },
    {
      "epoch": 18.99081102676788,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 47534
    },
    {
      "epoch": 18.991210547343186,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 47535
    },
    {
      "epoch": 18.991610067918497,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0126,
      "step": 47536
    },
    {
      "epoch": 18.992009588493808,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 47537
    },
    {
      "epoch": 18.99240910906912,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 47538
    },
    {
      "epoch": 18.992808629644426,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0286,
      "step": 47539
    },
    {
      "epoch": 18.993208150219736,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 47540
    },
    {
      "epoch": 18.993607670795047,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 47541
    },
    {
      "epoch": 18.994007191370354,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 47542
    },
    {
      "epoch": 18.994406711945665,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.027,
      "step": 47543
    },
    {
      "epoch": 18.994806232520975,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 47544
    },
    {
      "epoch": 18.995205753096286,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 47545
    },
    {
      "epoch": 18.995605273671593,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0318,
      "step": 47546
    },
    {
      "epoch": 18.996004794246904,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0336,
      "step": 47547
    },
    {
      "epoch": 18.996404314822215,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 47548
    },
    {
      "epoch": 18.99680383539752,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 47549
    },
    {
      "epoch": 18.997203355972832,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0005,
      "loss": 1.0108,
      "step": 47550
    },
    {
      "epoch": 18.997602876548143,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 47551
    },
    {
      "epoch": 18.99800239712345,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 47552
    },
    {
      "epoch": 18.99840191769876,
      "grad_norm": 0.140625,
      "learning_rate": 0.0005,
      "loss": 1.0374,
      "step": 47553
    },
    {
      "epoch": 18.99880143827407,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 47554
    },
    {
      "epoch": 18.999200958849382,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.039,
      "step": 47555
    },
    {
      "epoch": 18.99960047942469,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 47556
    },
    {
      "epoch": 19.0,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0122,
      "step": 47557
    },
    {
      "epoch": 19.00039952057531,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 47558
    },
    {
      "epoch": 19.000799041150618,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 47559
    },
    {
      "epoch": 19.00119856172593,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 47560
    },
    {
      "epoch": 19.00159808230124,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 47561
    },
    {
      "epoch": 19.00199760287655,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0315,
      "step": 47562
    },
    {
      "epoch": 19.002397123451857,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0263,
      "step": 47563
    },
    {
      "epoch": 19.002796644027168,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 47564
    },
    {
      "epoch": 19.00319616460248,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 47565
    },
    {
      "epoch": 19.003595685177785,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 47566
    },
    {
      "epoch": 19.003995205753096,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 47567
    },
    {
      "epoch": 19.004394726328407,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 47568
    },
    {
      "epoch": 19.004794246903714,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0128,
      "step": 47569
    },
    {
      "epoch": 19.005193767479025,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 47570
    },
    {
      "epoch": 19.005593288054335,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 47571
    },
    {
      "epoch": 19.005992808629646,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 47572
    },
    {
      "epoch": 19.006392329204953,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 47573
    },
    {
      "epoch": 19.006791849780264,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 47574
    },
    {
      "epoch": 19.007191370355574,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 47575
    },
    {
      "epoch": 19.00759089093088,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 47576
    },
    {
      "epoch": 19.007990411506192,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 47577
    },
    {
      "epoch": 19.008389932081503,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 47578
    },
    {
      "epoch": 19.008789452656814,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 47579
    },
    {
      "epoch": 19.00918897323212,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0324,
      "step": 47580
    },
    {
      "epoch": 19.00958849380743,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 47581
    },
    {
      "epoch": 19.009988014382742,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0191,
      "step": 47582
    },
    {
      "epoch": 19.01038753495805,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 47583
    },
    {
      "epoch": 19.01078705553336,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 47584
    },
    {
      "epoch": 19.01118657610867,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 47585
    },
    {
      "epoch": 19.011586096683978,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 47586
    },
    {
      "epoch": 19.01198561725929,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 47587
    },
    {
      "epoch": 19.0123851378346,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 47588
    },
    {
      "epoch": 19.01278465840991,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 47589
    },
    {
      "epoch": 19.013184178985217,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0208,
      "step": 47590
    },
    {
      "epoch": 19.013583699560527,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 47591
    },
    {
      "epoch": 19.013983220135838,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 47592
    },
    {
      "epoch": 19.014382740711145,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 47593
    },
    {
      "epoch": 19.014782261286456,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0166,
      "step": 47594
    },
    {
      "epoch": 19.015181781861767,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 47595
    },
    {
      "epoch": 19.015581302437077,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 47596
    },
    {
      "epoch": 19.015980823012384,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 47597
    },
    {
      "epoch": 19.016380343587695,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0005,
      "loss": 1.0285,
      "step": 47598
    },
    {
      "epoch": 19.016779864163006,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 47599
    },
    {
      "epoch": 19.017179384738313,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 47600
    },
    {
      "epoch": 19.017578905313623,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0135,
      "step": 47601
    },
    {
      "epoch": 19.017978425888934,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0087,
      "step": 47602
    },
    {
      "epoch": 19.01837794646424,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 47603
    },
    {
      "epoch": 19.018777467039552,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 47604
    },
    {
      "epoch": 19.019176987614863,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 47605
    },
    {
      "epoch": 19.019576508190173,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 47606
    },
    {
      "epoch": 19.01997602876548,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 47607
    },
    {
      "epoch": 19.02037554934079,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 47608
    },
    {
      "epoch": 19.020775069916102,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 47609
    },
    {
      "epoch": 19.02117459049141,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 47610
    },
    {
      "epoch": 19.02157411106672,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 47611
    },
    {
      "epoch": 19.02197363164203,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 47612
    },
    {
      "epoch": 19.02237315221734,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 47613
    },
    {
      "epoch": 19.022772672792648,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 47614
    },
    {
      "epoch": 19.02317219336796,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 47615
    },
    {
      "epoch": 19.02357171394327,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.0254,
      "step": 47616
    },
    {
      "epoch": 19.023971234518577,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 47617
    },
    {
      "epoch": 19.024370755093887,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 47618
    },
    {
      "epoch": 19.024770275669198,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 47619
    },
    {
      "epoch": 19.025169796244505,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 47620
    },
    {
      "epoch": 19.025569316819816,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0332,
      "step": 47621
    },
    {
      "epoch": 19.025968837395126,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0249,
      "step": 47622
    },
    {
      "epoch": 19.026368357970437,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 47623
    },
    {
      "epoch": 19.026767878545744,
      "grad_norm": 0.050048828125,
      "learning_rate": 0.0005,
      "loss": 1.0356,
      "step": 47624
    },
    {
      "epoch": 19.027167399121055,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 47625
    },
    {
      "epoch": 19.027566919696365,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0117,
      "step": 47626
    },
    {
      "epoch": 19.027966440271673,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 47627
    },
    {
      "epoch": 19.028365960846983,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 47628
    },
    {
      "epoch": 19.028765481422294,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 47629
    },
    {
      "epoch": 19.029165001997605,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0186,
      "step": 47630
    },
    {
      "epoch": 19.02956452257291,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.0347,
      "step": 47631
    },
    {
      "epoch": 19.029964043148222,
      "grad_norm": 0.0517578125,
      "learning_rate": 0.0005,
      "loss": 1.0135,
      "step": 47632
    },
    {
      "epoch": 19.030363563723533,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 47633
    },
    {
      "epoch": 19.03076308429884,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0005,
      "loss": 1.0145,
      "step": 47634
    },
    {
      "epoch": 19.03116260487415,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0169,
      "step": 47635
    },
    {
      "epoch": 19.03156212544946,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0413,
      "step": 47636
    },
    {
      "epoch": 19.03196164602477,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0243,
      "step": 47637
    },
    {
      "epoch": 19.03236116660008,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0199,
      "step": 47638
    },
    {
      "epoch": 19.03276068717539,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 47639
    },
    {
      "epoch": 19.0331602077507,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0281,
      "step": 47640
    },
    {
      "epoch": 19.033559728326008,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0322,
      "step": 47641
    },
    {
      "epoch": 19.03395924890132,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 47642
    },
    {
      "epoch": 19.03435876947663,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 47643
    },
    {
      "epoch": 19.034758290051936,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 47644
    },
    {
      "epoch": 19.035157810627247,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 47645
    },
    {
      "epoch": 19.035557331202558,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 47646
    },
    {
      "epoch": 19.03595685177787,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 47647
    },
    {
      "epoch": 19.036356372353175,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 47648
    },
    {
      "epoch": 19.036755892928486,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0178,
      "step": 47649
    },
    {
      "epoch": 19.037155413503797,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 47650
    },
    {
      "epoch": 19.037554934079104,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 47651
    },
    {
      "epoch": 19.037954454654415,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 47652
    },
    {
      "epoch": 19.038353975229725,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0334,
      "step": 47653
    },
    {
      "epoch": 19.038753495805032,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 47654
    },
    {
      "epoch": 19.039153016380343,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 47655
    },
    {
      "epoch": 19.039552536955654,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 47656
    },
    {
      "epoch": 19.039952057530964,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 47657
    },
    {
      "epoch": 19.04035157810627,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 47658
    },
    {
      "epoch": 19.040751098681582,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 47659
    },
    {
      "epoch": 19.041150619256893,
      "grad_norm": 0.03955078125,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 47660
    },
    {
      "epoch": 19.0415501398322,
      "grad_norm": 0.09375,
      "learning_rate": 0.0005,
      "loss": 1.0236,
      "step": 47661
    },
    {
      "epoch": 19.04194966040751,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0222,
      "step": 47662
    },
    {
      "epoch": 19.04234918098282,
      "grad_norm": 0.109375,
      "learning_rate": 0.0005,
      "loss": 0.9993,
      "step": 47663
    },
    {
      "epoch": 19.04274870155813,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0233,
      "step": 47664
    },
    {
      "epoch": 19.04314822213344,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0219,
      "step": 47665
    },
    {
      "epoch": 19.04354774270875,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 47666
    },
    {
      "epoch": 19.04394726328406,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 47667
    },
    {
      "epoch": 19.044346783859368,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 47668
    },
    {
      "epoch": 19.04474630443468,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0252,
      "step": 47669
    },
    {
      "epoch": 19.04514582500999,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 47670
    },
    {
      "epoch": 19.045545345585296,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0311,
      "step": 47671
    },
    {
      "epoch": 19.045944866160607,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 47672
    },
    {
      "epoch": 19.046344386735917,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 47673
    },
    {
      "epoch": 19.046743907311228,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 47674
    },
    {
      "epoch": 19.047143427886535,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0293,
      "step": 47675
    },
    {
      "epoch": 19.047542948461846,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 47676
    },
    {
      "epoch": 19.047942469037157,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0164,
      "step": 47677
    },
    {
      "epoch": 19.048341989612464,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 47678
    },
    {
      "epoch": 19.048741510187774,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0355,
      "step": 47679
    },
    {
      "epoch": 19.049141030763085,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 47680
    },
    {
      "epoch": 19.049540551338392,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 47681
    },
    {
      "epoch": 19.049940071913703,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0271,
      "step": 47682
    },
    {
      "epoch": 19.050339592489014,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0005,
      "loss": 1.0308,
      "step": 47683
    },
    {
      "epoch": 19.050739113064324,
      "grad_norm": 0.056396484375,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 47684
    },
    {
      "epoch": 19.05113863363963,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0242,
      "step": 47685
    },
    {
      "epoch": 19.051538154214942,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 47686
    },
    {
      "epoch": 19.051937674790253,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 47687
    },
    {
      "epoch": 19.05233719536556,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 47688
    },
    {
      "epoch": 19.05273671594087,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 47689
    },
    {
      "epoch": 19.05313623651618,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 47690
    },
    {
      "epoch": 19.053535757091492,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 47691
    },
    {
      "epoch": 19.0539352776668,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0005,
      "loss": 1.0171,
      "step": 47692
    },
    {
      "epoch": 19.05433479824211,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 47693
    },
    {
      "epoch": 19.05473431881742,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 47694
    },
    {
      "epoch": 19.055133839392727,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 47695
    },
    {
      "epoch": 19.055533359968038,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0307,
      "step": 47696
    },
    {
      "epoch": 19.05593288054335,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 47697
    },
    {
      "epoch": 19.056332401118656,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 47698
    },
    {
      "epoch": 19.056731921693967,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 47699
    },
    {
      "epoch": 19.057131442269277,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 47700
    },
    {
      "epoch": 19.057530962844588,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0005,
      "loss": 1.0228,
      "step": 47701
    },
    {
      "epoch": 19.057930483419895,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 47702
    },
    {
      "epoch": 19.058330003995206,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 47703
    },
    {
      "epoch": 19.058729524570516,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0207,
      "step": 47704
    },
    {
      "epoch": 19.059129045145824,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 47705
    },
    {
      "epoch": 19.059528565721134,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 47706
    },
    {
      "epoch": 19.059928086296445,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0165,
      "step": 47707
    },
    {
      "epoch": 19.060327606871756,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 47708
    },
    {
      "epoch": 19.060727127447063,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0005,
      "loss": 1.017,
      "step": 47709
    },
    {
      "epoch": 19.061126648022373,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 47710
    },
    {
      "epoch": 19.061526168597684,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 47711
    },
    {
      "epoch": 19.06192568917299,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0118,
      "step": 47712
    },
    {
      "epoch": 19.062325209748302,
      "grad_norm": 0.040771484375,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 47713
    },
    {
      "epoch": 19.062724730323612,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0288,
      "step": 47714
    },
    {
      "epoch": 19.06312425089892,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 47715
    },
    {
      "epoch": 19.06352377147423,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0287,
      "step": 47716
    },
    {
      "epoch": 19.06392329204954,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0272,
      "step": 47717
    },
    {
      "epoch": 19.06432281262485,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 47718
    },
    {
      "epoch": 19.06472233320016,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 47719
    },
    {
      "epoch": 19.06512185377547,
      "grad_norm": 0.049072265625,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 47720
    },
    {
      "epoch": 19.06552137435078,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 47721
    },
    {
      "epoch": 19.065920894926087,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 47722
    },
    {
      "epoch": 19.066320415501398,
      "grad_norm": 0.036865234375,
      "learning_rate": 0.0005,
      "loss": 1.0358,
      "step": 47723
    },
    {
      "epoch": 19.06671993607671,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 47724
    },
    {
      "epoch": 19.06711945665202,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0167,
      "step": 47725
    },
    {
      "epoch": 19.067518977227326,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 47726
    },
    {
      "epoch": 19.067918497802637,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.035,
      "step": 47727
    },
    {
      "epoch": 19.068318018377948,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 47728
    },
    {
      "epoch": 19.068717538953255,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 47729
    },
    {
      "epoch": 19.069117059528566,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0241,
      "step": 47730
    },
    {
      "epoch": 19.069516580103876,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0435,
      "step": 47731
    },
    {
      "epoch": 19.069916100679183,
      "grad_norm": 0.052978515625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 47732
    },
    {
      "epoch": 19.070315621254494,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 47733
    },
    {
      "epoch": 19.070715141829805,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 47734
    },
    {
      "epoch": 19.071114662405115,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0202,
      "step": 47735
    },
    {
      "epoch": 19.071514182980422,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 47736
    },
    {
      "epoch": 19.071913703555733,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.016,
      "step": 47737
    },
    {
      "epoch": 19.072313224131044,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 47738
    },
    {
      "epoch": 19.07271274470635,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 47739
    },
    {
      "epoch": 19.07311226528166,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0162,
      "step": 47740
    },
    {
      "epoch": 19.073511785856972,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 47741
    },
    {
      "epoch": 19.073911306432283,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0214,
      "step": 47742
    },
    {
      "epoch": 19.07431082700759,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 47743
    },
    {
      "epoch": 19.0747103475829,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0005,
      "loss": 1.025,
      "step": 47744
    },
    {
      "epoch": 19.07510986815821,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 47745
    },
    {
      "epoch": 19.07550938873352,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0205,
      "step": 47746
    },
    {
      "epoch": 19.07590890930883,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 47747
    },
    {
      "epoch": 19.07630842988414,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 47748
    },
    {
      "epoch": 19.076707950459447,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0005,
      "loss": 1.0296,
      "step": 47749
    },
    {
      "epoch": 19.077107471034758,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 47750
    },
    {
      "epoch": 19.07750699161007,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.022,
      "step": 47751
    },
    {
      "epoch": 19.07790651218538,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 47752
    },
    {
      "epoch": 19.078306032760686,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 47753
    },
    {
      "epoch": 19.078705553335997,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0005,
      "loss": 1.033,
      "step": 47754
    },
    {
      "epoch": 19.079105073911307,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.032,
      "step": 47755
    },
    {
      "epoch": 19.079504594486615,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 47756
    },
    {
      "epoch": 19.079904115061925,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 47757
    },
    {
      "epoch": 19.080303635637236,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 47758
    },
    {
      "epoch": 19.080703156212547,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 47759
    },
    {
      "epoch": 19.081102676787854,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 47760
    },
    {
      "epoch": 19.081502197363164,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0108,
      "step": 47761
    },
    {
      "epoch": 19.081901717938475,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 47762
    },
    {
      "epoch": 19.082301238513782,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 47763
    },
    {
      "epoch": 19.082700759089093,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 47764
    },
    {
      "epoch": 19.083100279664404,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0005,
      "loss": 1.0261,
      "step": 47765
    },
    {
      "epoch": 19.08349980023971,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 47766
    },
    {
      "epoch": 19.08389932081502,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 47767
    },
    {
      "epoch": 19.084298841390332,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0304,
      "step": 47768
    },
    {
      "epoch": 19.084698361965643,
      "grad_norm": 0.07177734375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 47769
    },
    {
      "epoch": 19.08509788254095,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0339,
      "step": 47770
    },
    {
      "epoch": 19.08549740311626,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 47771
    },
    {
      "epoch": 19.08589692369157,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0177,
      "step": 47772
    },
    {
      "epoch": 19.08629644426688,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 47773
    },
    {
      "epoch": 19.08669596484219,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 47774
    },
    {
      "epoch": 19.0870954854175,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0005,
      "loss": 1.0297,
      "step": 47775
    },
    {
      "epoch": 19.08749500599281,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 47776
    },
    {
      "epoch": 19.087894526568117,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 47777
    },
    {
      "epoch": 19.088294047143428,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0159,
      "step": 47778
    },
    {
      "epoch": 19.08869356771874,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0005,
      "loss": 1.0335,
      "step": 47779
    },
    {
      "epoch": 19.089093088294046,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 47780
    },
    {
      "epoch": 19.089492608869357,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 47781
    },
    {
      "epoch": 19.089892129444667,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 47782
    },
    {
      "epoch": 19.090291650019974,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 47783
    },
    {
      "epoch": 19.090691170595285,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 47784
    },
    {
      "epoch": 19.091090691170596,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0156,
      "step": 47785
    },
    {
      "epoch": 19.091490211745906,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0301,
      "step": 47786
    },
    {
      "epoch": 19.091889732321214,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0383,
      "step": 47787
    },
    {
      "epoch": 19.092289252896524,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 47788
    },
    {
      "epoch": 19.092688773471835,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 47789
    },
    {
      "epoch": 19.093088294047142,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0147,
      "step": 47790
    },
    {
      "epoch": 19.093487814622453,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 47791
    },
    {
      "epoch": 19.093887335197763,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0005,
      "loss": 1.0251,
      "step": 47792
    },
    {
      "epoch": 19.094286855773074,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0135,
      "step": 47793
    },
    {
      "epoch": 19.09468637634838,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0333,
      "step": 47794
    },
    {
      "epoch": 19.095085896923692,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.029,
      "step": 47795
    },
    {
      "epoch": 19.095485417499003,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 47796
    },
    {
      "epoch": 19.09588493807431,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 47797
    },
    {
      "epoch": 19.09628445864962,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0005,
      "loss": 1.024,
      "step": 47798
    },
    {
      "epoch": 19.09668397922493,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0045,
      "step": 47799
    },
    {
      "epoch": 19.097083499800238,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 0.9984,
      "step": 47800
    },
    {
      "epoch": 19.09748302037555,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 47801
    },
    {
      "epoch": 19.09788254095086,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0005,
      "loss": 1.0346,
      "step": 47802
    },
    {
      "epoch": 19.09828206152617,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 47803
    },
    {
      "epoch": 19.098681582101477,
      "grad_norm": 0.05322265625,
      "learning_rate": 0.0005,
      "loss": 1.0317,
      "step": 47804
    },
    {
      "epoch": 19.099081102676788,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 47805
    },
    {
      "epoch": 19.0994806232521,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0005,
      "loss": 1.0316,
      "step": 47806
    },
    {
      "epoch": 19.099880143827406,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 47807
    },
    {
      "epoch": 19.100279664402716,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 47808
    },
    {
      "epoch": 19.100679184978027,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 47809
    },
    {
      "epoch": 19.101078705553338,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0212,
      "step": 47810
    },
    {
      "epoch": 19.101478226128645,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 47811
    },
    {
      "epoch": 19.101877746703956,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 47812
    },
    {
      "epoch": 19.102277267279266,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0005,
      "loss": 1.0323,
      "step": 47813
    },
    {
      "epoch": 19.102676787854573,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 47814
    },
    {
      "epoch": 19.103076308429884,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0158,
      "step": 47815
    },
    {
      "epoch": 19.103475829005195,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0245,
      "step": 47816
    },
    {
      "epoch": 19.103875349580502,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0111,
      "step": 47817
    },
    {
      "epoch": 19.104274870155812,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0005,
      "loss": 1.034,
      "step": 47818
    },
    {
      "epoch": 19.104674390731123,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0244,
      "step": 47819
    },
    {
      "epoch": 19.105073911306434,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0005,
      "loss": 1.0213,
      "step": 47820
    },
    {
      "epoch": 19.10547343188174,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 47821
    },
    {
      "epoch": 19.10587295245705,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 47822
    },
    {
      "epoch": 19.106272473032362,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 47823
    },
    {
      "epoch": 19.10667199360767,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.013,
      "step": 47824
    },
    {
      "epoch": 19.10707151418298,
      "grad_norm": 0.0537109375,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 47825
    },
    {
      "epoch": 19.10747103475829,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0005,
      "loss": 1.0197,
      "step": 47826
    },
    {
      "epoch": 19.107870555333598,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 47827
    },
    {
      "epoch": 19.10827007590891,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0005,
      "loss": 1.0382,
      "step": 47828
    },
    {
      "epoch": 19.10866959648422,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0387,
      "step": 47829
    },
    {
      "epoch": 19.10906911705953,
      "grad_norm": 0.051025390625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 47830
    },
    {
      "epoch": 19.109468637634837,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 47831
    },
    {
      "epoch": 19.109868158210148,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0345,
      "step": 47832
    },
    {
      "epoch": 19.11026767878546,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.023,
      "step": 47833
    },
    {
      "epoch": 19.110667199360766,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 47834
    },
    {
      "epoch": 19.111066719936076,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 47835
    },
    {
      "epoch": 19.111466240511387,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0005,
      "loss": 1.014,
      "step": 47836
    },
    {
      "epoch": 19.111865761086698,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 47837
    },
    {
      "epoch": 19.112265281662005,
      "grad_norm": 0.041015625,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 47838
    },
    {
      "epoch": 19.112664802237315,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0005,
      "loss": 1.0319,
      "step": 47839
    },
    {
      "epoch": 19.113064322812626,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 47840
    },
    {
      "epoch": 19.113463843387933,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0204,
      "step": 47841
    },
    {
      "epoch": 19.113863363963244,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0234,
      "step": 47842
    },
    {
      "epoch": 19.114262884538554,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 47843
    },
    {
      "epoch": 19.11466240511386,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 47844
    },
    {
      "epoch": 19.115061925689172,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0116,
      "step": 47845
    },
    {
      "epoch": 19.115461446264483,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 47846
    },
    {
      "epoch": 19.115860966839794,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 47847
    },
    {
      "epoch": 19.1162604874151,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0354,
      "step": 47848
    },
    {
      "epoch": 19.11666000799041,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0133,
      "step": 47849
    },
    {
      "epoch": 19.117059528565722,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0157,
      "step": 47850
    },
    {
      "epoch": 19.11745904914103,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0005,
      "loss": 1.0091,
      "step": 47851
    },
    {
      "epoch": 19.11785856971634,
      "grad_norm": 0.04638671875,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 47852
    },
    {
      "epoch": 19.11825809029165,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0107,
      "step": 47853
    },
    {
      "epoch": 19.11865761086696,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 47854
    },
    {
      "epoch": 19.11905713144227,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0247,
      "step": 47855
    },
    {
      "epoch": 19.11945665201758,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.026,
      "step": 47856
    },
    {
      "epoch": 19.11985617259289,
      "grad_norm": 0.044677734375,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 47857
    },
    {
      "epoch": 19.120255693168197,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 47858
    },
    {
      "epoch": 19.120655213743508,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0184,
      "step": 47859
    },
    {
      "epoch": 19.121054734318818,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 47860
    },
    {
      "epoch": 19.121454254894125,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 47861
    },
    {
      "epoch": 19.121853775469436,
      "grad_norm": 0.043212890625,
      "learning_rate": 0.0005,
      "loss": 1.0259,
      "step": 47862
    },
    {
      "epoch": 19.122253296044747,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0338,
      "step": 47863
    },
    {
      "epoch": 19.122652816620057,
      "grad_norm": 0.0625,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 47864
    },
    {
      "epoch": 19.123052337195364,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0321,
      "step": 47865
    },
    {
      "epoch": 19.123451857770675,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0099,
      "step": 47866
    },
    {
      "epoch": 19.123851378345986,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0206,
      "step": 47867
    },
    {
      "epoch": 19.124250898921293,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0182,
      "step": 47868
    },
    {
      "epoch": 19.124650419496604,
      "grad_norm": 0.05859375,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 47869
    },
    {
      "epoch": 19.125049940071914,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 47870
    },
    {
      "epoch": 19.125449460647225,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0005,
      "loss": 1.0239,
      "step": 47871
    },
    {
      "epoch": 19.125848981222532,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0005,
      "loss": 1.0298,
      "step": 47872
    },
    {
      "epoch": 19.126248501797843,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0194,
      "step": 47873
    },
    {
      "epoch": 19.126648022373153,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 47874
    },
    {
      "epoch": 19.12704754294846,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 47875
    },
    {
      "epoch": 19.12744706352377,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 47876
    },
    {
      "epoch": 19.127846584099082,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 47877
    },
    {
      "epoch": 19.12824610467439,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 47878
    },
    {
      "epoch": 19.1286456252497,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0005,
      "loss": 1.0155,
      "step": 47879
    },
    {
      "epoch": 19.12904514582501,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0005,
      "loss": 1.0119,
      "step": 47880
    },
    {
      "epoch": 19.12944466640032,
      "grad_norm": 0.044189453125,
      "learning_rate": 0.0005,
      "loss": 1.0196,
      "step": 47881
    },
    {
      "epoch": 19.129844186975628,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0005,
      "loss": 1.031,
      "step": 47882
    },
    {
      "epoch": 19.13024370755094,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 47883
    },
    {
      "epoch": 19.13064322812625,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0295,
      "step": 47884
    },
    {
      "epoch": 19.131042748701557,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0005,
      "loss": 1.0109,
      "step": 47885
    },
    {
      "epoch": 19.131442269276867,
      "grad_norm": 0.045654296875,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 47886
    },
    {
      "epoch": 19.131841789852178,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0129,
      "step": 47887
    },
    {
      "epoch": 19.13224131042749,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0005,
      "loss": 1.0195,
      "step": 47888
    },
    {
      "epoch": 19.132640831002796,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0005,
      "loss": 1.0113,
      "step": 47889
    },
    {
      "epoch": 19.133040351578106,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 47890
    },
    {
      "epoch": 19.133439872153417,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 47891
    },
    {
      "epoch": 19.133839392728724,
      "grad_norm": 0.078125,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 47892
    },
    {
      "epoch": 19.134238913304035,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0005,
      "loss": 1.0125,
      "step": 47893
    },
    {
      "epoch": 19.134638433879346,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0005,
      "loss": 1.0127,
      "step": 47894
    },
    {
      "epoch": 19.135037954454653,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0273,
      "step": 47895
    },
    {
      "epoch": 19.135437475029963,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0005,
      "loss": 1.0395,
      "step": 47896
    },
    {
      "epoch": 19.135836995605274,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0255,
      "step": 47897
    },
    {
      "epoch": 19.136236516180585,
      "grad_norm": 0.0751953125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 47898
    },
    {
      "epoch": 19.136636036755892,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0217,
      "step": 47899
    },
    {
      "epoch": 19.137035557331203,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0209,
      "step": 47900
    },
    {
      "epoch": 19.137435077906513,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0005,
      "loss": 1.0181,
      "step": 47901
    },
    {
      "epoch": 19.13783459848182,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0108,
      "step": 47902
    },
    {
      "epoch": 19.13823411905713,
      "grad_norm": 0.04443359375,
      "learning_rate": 0.0005,
      "loss": 1.0141,
      "step": 47903
    },
    {
      "epoch": 19.13863363963244,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0005,
      "loss": 1.028,
      "step": 47904
    },
    {
      "epoch": 19.139033160207752,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0005,
      "loss": 1.0146,
      "step": 47905
    },
    {
      "epoch": 19.13943268078306,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0303,
      "step": 47906
    },
    {
      "epoch": 19.13983220135837,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0314,
      "step": 47907
    },
    {
      "epoch": 19.14023172193368,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0238,
      "step": 47908
    },
    {
      "epoch": 19.140631242508988,
      "grad_norm": 0.049560546875,
      "learning_rate": 0.0005,
      "loss": 1.0119,
      "step": 47909
    },
    {
      "epoch": 19.1410307630843,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0005,
      "loss": 1.01,
      "step": 47910
    },
    {
      "epoch": 19.14143028365961,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0391,
      "step": 47911
    },
    {
      "epoch": 19.141829804234916,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 47912
    },
    {
      "epoch": 19.142229324810227,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0005,
      "loss": 1.0142,
      "step": 47913
    },
    {
      "epoch": 19.142628845385538,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0005,
      "loss": 1.0327,
      "step": 47914
    },
    {
      "epoch": 19.14302836596085,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0005,
      "loss": 1.0309,
      "step": 47915
    },
    {
      "epoch": 19.143427886536156,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 47916
    },
    {
      "epoch": 19.143827407111466,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0005,
      "loss": 1.0176,
      "step": 47917
    },
    {
      "epoch": 19.144226927686777,
      "grad_norm": 0.056640625,
      "learning_rate": 0.0005,
      "loss": 1.0185,
      "step": 47918
    },
    {
      "epoch": 19.144626448262084,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 47919
    },
    {
      "epoch": 19.145025968837395,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0306,
      "step": 47920
    },
    {
      "epoch": 19.145425489412705,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0097,
      "step": 47921
    },
    {
      "epoch": 19.145825009988016,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 47922
    },
    {
      "epoch": 19.146224530563323,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 47923
    },
    {
      "epoch": 19.146624051138634,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0005,
      "loss": 1.0274,
      "step": 47924
    },
    {
      "epoch": 19.147023571713945,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0005,
      "loss": 1.0394,
      "step": 47925
    },
    {
      "epoch": 19.14742309228925,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0005,
      "loss": 1.0248,
      "step": 47926
    },
    {
      "epoch": 19.147822612864562,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0005,
      "loss": 1.0262,
      "step": 47927
    },
    {
      "epoch": 19.148222133439873,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 47928
    },
    {
      "epoch": 19.14862165401518,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0264,
      "step": 47929
    },
    {
      "epoch": 19.14902117459049,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0005,
      "loss": 1.0283,
      "step": 47930
    },
    {
      "epoch": 19.1494206951658,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0005,
      "loss": 1.0188,
      "step": 47931
    },
    {
      "epoch": 19.149820215741112,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0005,
      "loss": 1.0299,
      "step": 47932
    },
    {
      "epoch": 19.15021973631642,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0005,
      "loss": 1.0227,
      "step": 47933
    },
    {
      "epoch": 19.15061925689173,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0005,
      "loss": 1.0269,
      "step": 47934
    },
    {
      "epoch": 19.15101877746704,
      "grad_norm": 0.046630859375,
      "learning_rate": 0.0005,
      "loss": 1.0122,
      "step": 47935
    },
    {
      "epoch": 19.151418298042348,
      "grad_norm": 0.07373046875,
      "learning_rate": 0.0005,
      "loss": 1.0126,
      "step": 47936
    },
    {
      "epoch": 19.15181781861766,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0005,
      "loss": 1.0266,
      "step": 47937
    },
    {
      "epoch": 19.15221733919297,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0005,
      "loss": 1.0118,
      "step": 47938
    },
    {
      "epoch": 19.15261685976828,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0005,
      "loss": 1.0122,
      "step": 47939
    },
    {
      "epoch": 19.153016380343587,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 47940
    },
    {
      "epoch": 19.153415900918898,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0431,
      "step": 47941
    },
    {
      "epoch": 19.15381542149421,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0005,
      "loss": 1.0174,
      "step": 47942
    },
    {
      "epoch": 19.154214942069515,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 47943
    },
    {
      "epoch": 19.154614462644826,
      "grad_norm": 0.061767578125,
      "learning_rate": 0.0005,
      "loss": 1.0268,
      "step": 47944
    },
    {
      "epoch": 19.155013983220137,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 47945
    },
    {
      "epoch": 19.155413503795444,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 47946
    },
    {
      "epoch": 19.155813024370755,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0005,
      "loss": 1.0211,
      "step": 47947
    },
    {
      "epoch": 19.156212544946065,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0005,
      "loss": 1.0279,
      "step": 47948
    },
    {
      "epoch": 19.156612065521376,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 47949
    },
    {
      "epoch": 19.157011586096683,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.02,
      "step": 47950
    },
    {
      "epoch": 19.157411106671994,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0005,
      "loss": 1.0235,
      "step": 47951
    },
    {
      "epoch": 19.157810627247304,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 47952
    },
    {
      "epoch": 19.15821014782261,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0351,
      "step": 47953
    },
    {
      "epoch": 19.158609668397922,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 47954
    },
    {
      "epoch": 19.159009188973233,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0005,
      "loss": 1.0265,
      "step": 47955
    },
    {
      "epoch": 19.159408709548543,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 47956
    },
    {
      "epoch": 19.15980823012385,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0005,
      "loss": 1.0352,
      "step": 47957
    },
    {
      "epoch": 19.16020775069916,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0005,
      "loss": 1.0246,
      "step": 47958
    },
    {
      "epoch": 19.160607271274472,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0005,
      "loss": 1.0201,
      "step": 47959
    },
    {
      "epoch": 19.16100679184978,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0005,
      "loss": 1.0153,
      "step": 47960
    },
    {
      "epoch": 19.16140631242509,
      "grad_norm": 0.06103515625,
      "learning_rate": 0.0005,
      "loss": 1.0256,
      "step": 47961
    },
    {
      "epoch": 19.1618058330004,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0005,
      "loss": 1.0368,
      "step": 47962
    },
    {
      "epoch": 19.162205353575708,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0005,
      "loss": 1.018,
      "step": 47963
    },
    {
      "epoch": 19.162604874151018,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0005,
      "loss": 1.0192,
      "step": 47964
    },
    {
      "epoch": 19.16300439472633,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0005,
      "loss": 1.0292,
      "step": 47965
    },
    {
      "epoch": 19.16340391530164,
      "grad_norm": 0.053466796875,
      "learning_rate": 0.0005,
      "loss": 1.0221,
      "step": 47966
    },
    {
      "epoch": 19.163803435876947,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0005,
      "loss": 1.0138,
      "step": 47967
    },
    {
      "epoch": 19.164202956452257,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0005,
      "loss": 1.0137,
      "step": 47968
    },
    {
      "epoch": 19.164602477027568,
      "grad_norm": 0.0419921875,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 47969
    },
    {
      "epoch": 19.165001997602875,
      "grad_norm": 0.04931640625,
      "learning_rate": 0.0005,
      "loss": 1.0193,
      "step": 47970
    },
    {
      "epoch": 19.165401518178186,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0152,
      "step": 47971
    },
    {
      "epoch": 19.165801038753496,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0005,
      "loss": 1.0123,
      "step": 47972
    },
    {
      "epoch": 19.166200559328807,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0005,
      "loss": 1.0275,
      "step": 47973
    },
    {
      "epoch": 19.166600079904114,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0005,
      "loss": 1.0289,
      "step": 47974
    },
    {
      "epoch": 19.166999600479425,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0005,
      "loss": 1.0232,
      "step": 47975
    },
    {
      "epoch": 19.167399121054736,
      "grad_norm": 0.039306640625,
      "learning_rate": 0.0005,
      "loss": 1.0175,
      "step": 47976
    },
    {
      "epoch": 19.167798641630043,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0005,
      "loss": 1.0179,
      "step": 47977
    },
    {
      "epoch": 19.168198162205353,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0005,
      "loss": 1.019,
      "step": 47978
    },
    {
      "epoch": 19.168597682780664,
      "grad_norm": 0.046875,
      "learning_rate": 0.0005,
      "loss": 1.0278,
      "step": 47979
    },
    {
      "epoch": 19.16899720335597,
      "grad_norm": 0.04150390625,
      "learning_rate": 0.0005,
      "loss": 1.0198,
      "step": 47980
    },
    {
      "epoch": 19.169396723931282,
      "grad_norm": 0.060791015625,
      "learning_rate": 0.0005,
      "loss": 1.0172,
      "step": 47981
    },
    {
      "epoch": 19.169796244506593,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0005,
      "loss": 1.0237,
      "step": 47982
    },
    {
      "epoch": 19.170195765081903,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0005,
      "loss": 1.0095,
      "step": 47983
    },
    {
      "epoch": 19.17059528565721,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0005,
      "loss": 1.0224,
      "step": 47984
    },
    {
      "epoch": 19.17099480623252,
      "grad_norm": 0.048828125,
      "learning_rate": 0.0005,
      "loss": 1.0215,
      "step": 47985
    },
    {
      "epoch": 19.17139432680783,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0183,
      "step": 47986
    },
    {
      "epoch": 19.17179384738314,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0005,
      "loss": 1.0284,
      "step": 47987
    },
    {
      "epoch": 19.17219336795845,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0005,
      "loss": 1.0151,
      "step": 47988
    },
    {
      "epoch": 19.17259288853376,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0005,
      "loss": 1.0173,
      "step": 47989
    },
    {
      "epoch": 19.17299240910907,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0005,
      "loss": 1.0258,
      "step": 47990
    },
    {
      "epoch": 19.173391929684378,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0005,
      "loss": 1.0203,
      "step": 47991
    },
    {
      "epoch": 19.17379145025969,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0005,
      "loss": 1.0216,
      "step": 47992
    },
    {
      "epoch": 19.174190970835,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0005,
      "loss": 1.0291,
      "step": 47993
    },
    {
      "epoch": 19.174590491410306,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0005,
      "loss": 1.0225,
      "step": 47994
    },
    {
      "epoch": 19.174990011985617,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0005,
      "loss": 1.0226,
      "step": 47995
    },
    {
      "epoch": 19.175389532560928,
      "grad_norm": 0.05126953125,
      "learning_rate": 0.0005,
      "loss": 1.0223,
      "step": 47996
    },
    {
      "epoch": 19.175789053136235,
      "grad_norm": 0.05712890625,
      "learning_rate": 0.0005,
      "loss": 1.0253,
      "step": 47997
    },
    {
      "epoch": 19.176188573711546,
      "grad_norm": 0.05029296875,
      "learning_rate": 0.0005,
      "loss": 1.0331,
      "step": 47998
    },
    {
      "epoch": 19.176588094286856,
      "grad_norm": 0.045166015625,
      "learning_rate": 0.0005,
      "loss": 1.0086,
      "step": 47999
    },
    {
      "epoch": 19.176987614862167,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0005,
      "loss": 1.0231,
      "step": 48000
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 2503000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1000,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.4562293160320434e+19,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": null
}